Computer system and program

Abstract

To provide a technology to enable an utterance other than an utterance on a progressing present situation to be made and to improve amusement properties in utterance control related to the entertainment.SOLUTION: A computer system predicts prediction events that can occur in the entertainment at a given prediction timing on the basis of progress state data and generates a plurality of prediction events. The computer system sets an adoption prediction event out of the plurality of prediction events, and determines whether or not a prediction event of the adoption prediction event has occurred in the entertainment after the prediction timing. The computer system selects a prediction event to be mentioned on the basis of a determination result, and causes a live character 12 to make an utterance on the prediction event to be mentioned.SELECTED DRAWING: Figure 6

Description

[Technical field]

[0001] The present invention relates to computer systems and the like. [Background technology]

[0002] There is known a type of video game in which a commentary sound is output as a speech control according to the progress of the game, like a sports broadcast on television. For example, Patent Document 1 describes a technology for a game such as horse racing, which predicts and calculates the game development in advance and generates a plurality of commentary sounds corresponding to each predicted situation in advance, thereby realizing smooth commentary speech control that does not fall behind the ever-changing situation. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2006-87620 A Summary of the Invention [Problem to be solved by the invention]

[0004] However, in conventional games, automatic commentary and other speech control mainly focused on explaining the current situation of the game. This problem is not limited to video games in the sports genre, but is similar regardless of genre, such as card games, role-playing games, and simulation games.

[0005] In addition to games, the same problem was encountered when generating so-called "live commentary" videos that automatically add commentary to real-world sports broadcasts, etc. The same problem was also encountered in speech control, not limited to live commentary, but also to commentary, conversations and monologues of virtual characters who are game players, etc.

[0006] The problem that the present invention aims to solve is to provide a technology for controlling speech in entertainment that enables speech other than that about the current ongoing situation, thereby improving interest. [Means for solving the problem]

[0007] A first invention for solving the above problems is a computer system (e.g., server system 1100 in FIG. 1 ) that performs speech control to utter a topic related to a given entertainment in text and / or voice based on progress status data of the entertainment (e.g., game progress status data 707 in FIG. 11 ), comprising: A prediction means (e.g., the server processing unit 200s and the prediction unit 214 in FIG. 8, the prediction timing determination condition data 562 in FIG. 9, and step S30 in FIG. 13) for predicting a predicted event that may occur in the entertainment at a given prediction timing based on the progress status data; A determination means (e.g., the server processing unit 200s, the determination unit 216 in FIG. 8, step S52 in FIG. 13) for determining whether or not the predicted event has occurred in the entertainment after the predicted timing; The computer system includes an utterance control means (e.g., the server processing unit 200s in FIG. 8, the utterance control unit 220, the player utterance generation model data 536 in FIG. 10, the live utterance generation model data 542, step S88 in FIG. 14, and steps S90 to S94 in FIG. 16) which performs the utterance control based on the judgment result by the judgment means.

[0008] The computer system of the first invention predicts a predicted event that may occur in the future based on the progress status data of the entertainment, and controls speech based on the result of judging whether the predicted event has occurred (whether it has been realized in the entertainment). Therefore, it is possible to realize a retrospective speech about a predicted event predicted in the past. This makes it possible to make speech other than speech about the current situation that is currently progressing, and improves the entertainment value.

[0009] A second invention is a computer system in which the speech control means controls the speech content based on speech tendency data set for a given character (e.g., player's speech tendency data 537, commentary speech tendency data 543 in FIG. 10), thereby controlling the speech so that the character speaks.

[0010] The computer system of the second invention can make the character speak based on the speech tendency set for the character. The speech tendency is the way of speaking (choice of words, amount of words, etc.), the frequency of spontaneous speech even when not spoken to by the other party, the degree of talkativeness / silence, etc. Therefore, a user who reads the character's spoken text or hears the character's spoken voice can feel as if the character has a personality, further increasing the interest.

[0011] A third invention is a computer system in which the entertainment is content progressed in a given virtual world, and the progress data is data relating to the progress of the content in the virtual world (e.g., game progress data 707 in FIG. 11).

[0012] According to the third invention, for example, for content that progresses in a virtual world such as a video game, it is possible to realize speech that looks back on predicted events that were predicted at a past point in time.

[0013] A fourth invention is a computer system in which the entertainment is content that takes place in the real world, and the progress status data is data related to the progress of the content (for example, status data 22 in Figure 22).

[0014] According to the fourth invention, for example, it is possible to realize speech that looks back on predicted events that were predicted at a past point in time regarding content that is taking place in the real world, such as sports.

[0015] A fifth invention is a computer system in which the prediction means determines that the prediction timing has arrived when the progress of the entertainment satisfies a given scene condition (e.g., scene condition data 563 in Figure 9), and predicts the predicted event.

[0016] According to the fifth aspect of the present invention, the computer system can predict a predicted event using the occurrence of a certain scene or situation as a trigger.

[0017] A sixth invention is a computer system in which, in the above-mentioned computer system, there are a plurality of scene conditions, and a plurality of predicted event candidates that are candidates for the predicted event are set for each scene condition (for example, prediction model and scene condition data 770 and predicted event candidate 774 in Figure 25), and the prediction means predicts the predicted event based on the plurality of predicted event candidates that are set for the scene condition satisfied by the progress status.

[0018] According to the sixth aspect of the present invention, the computer system can predict a predicted event from among predicted event candidates that are set in advance for each scene condition.

[0019] A seventh invention is the computer system described above, wherein the prediction means predicts the predicted event based on the progress state when the scene condition is satisfied.

[0020] According to a seventh aspect of the present invention, the computer system can predict a predicted event based on the progress of the entertainment when a scene condition is satisfied.

[0021] An eighth invention is a computer system in which the prediction means predicts the predicted event based on the history of the progress up to the point where the scene condition is satisfied (for example, data describing the chronological progress stored in the game progress data 707 in Figure 11).

[0022] According to an eighth aspect of the present invention, the computer system can predict a predicted event based on the progress of the situation up to the point where a scene condition is satisfied.

[0023] A ninth invention is a computer system in which the scene conditions are conditions regarding a scene involving a character appearing in the entertainment, and the prediction means predicts the predicted event based on information about the character (e.g., character initial setting data 512 in Figure 9).

[0024] According to the ninth aspect of the present invention, the computer system can predict a predicted event based on the characters appearing at the predicted timing.

[0025] A tenth invention is a computer system in which the entertainment is game content played by a player, and the prediction means predicts the predicted event based on player information of the player (for example, past play performance data of user management data 600 in FIG. 9, player level 531 and character attribute setting 532 in FIG. 10, and game operation history data 754 in FIG. 18).

[0026] According to the tenth aspect, the computer system can predict a predicted event based on player information.

[0027] An eleventh aspect of the present invention is the computer system described above, wherein the prediction means predicts a degree of likelihood of occurrence of the predicted event.

[0028] According to the eleventh aspect of the present invention, the computer system can predict the degree of likelihood of a predicted event occurring.

[0029] A twelfth invention is a computer system according to the above-mentioned embodiment, wherein the prediction means predicts a plurality of the predicted events, and the speech control means performs the speech control to speak content relating to a predicted event that is determined by the determination means to have not occurred among the plurality of predicted events predicted by the prediction means.

[0030] According to the twelfth aspect of the present invention, the computer system can execute speech control to speak content related to a predicted event that did not occur in the entertainment after the predicted timing.

[0031] A thirteenth invention is a computer system wherein, in the above-mentioned computer system, the entertainment is game content played by a player, the prediction means predicts the predicted events, and the speech control means selects at least one predicted event from the predicted events determined by the determination means to have not occurred based on player information of the player, and performs the speech control to speak content related to the selected predicted event.

[0032] According to the thirteenth aspect of the present invention, the computer system can select at least one of the predicted circumstances that has not occurred after the prediction based on the player information, and execute speech control to speak the content of the selected predicted circumstances.

[0033] A fourteenth invention is a computer system wherein, in the above-mentioned computer system, the player information includes playing tendency information (e.g., computer-controlled player information 515 and computer-controlled player playing tendency information 516 in FIG. 9 ), the prediction means performs the prediction by including a tendency-matching predicted event that matches a playing tendency based on the playing tendency information in the predicted event to be predicted, and the utterance control means, when the tendency-matching predicted event is included in the predicted event determined by the determination means to have not occurred, selects the tendency-matching predicted event and performs the utterance control.

[0034] According to the fourteenth aspect of the present invention, when a tendency-matching prediction event is included in a prediction situation that has not occurred after prediction, the computer system can execute speech control to speak the content of this.

[0035] A fifteenth invention is a computer system in which the player information includes playing tendency information, the prediction means executes the prediction by including a tendency incompatible predicted event that does not match a playing tendency based on the playing tendency information in the predicted events to be predicted, and the utterance control means, when the tendency incompatible predicted event is included in the predicted events determined by the determination means to have not occurred, selects the tendency incompatible predicted event and performs the utterance control.

[0036] According to the fifteenth aspect of the present invention, the computer system can execute speech control to speak the contents of a trend-incompatible prediction event when the prediction circumstances that have not occurred after prediction include the trend-incompatible prediction event.

[0037] A sixteenth invention is a computer system in which, in the above-mentioned computer system, the speech control means selects a predicted event to be spoken about based on past events occurring in the entertainment from the predicted events that have been determined by the judgment means to have not occurred, and performs the speech control.

[0038] According to the sixteenth aspect of the present invention, the computer system can select one predicted event based on a past event occurring in entertainment and perform speech control.

[0039] A 17th invention is a computer system in which, in the above-mentioned computer system, the prediction means predicts a degree of occurrence likelihood (e.g., predicted probability 43 in FIG. 4) for each of the predicted events, and the speech control means selects a predicted event to be spoken about based on the degree of occurrence likelihood for each of the predicted events that have been determined by the determination means to have not occurred, and performs the speech control.

[0040] According to the seventeenth aspect of the present invention, the computer system can select predicted events to be targets of speech control, based on the degree of occurrence probability of each predicted event that is determined not to have occurred in the entertainment after the predicted timing.

[0041] An eighteenth aspect of the present invention is the above-mentioned computer system, wherein the speech control means comprises: Performing the speech control so that a given character speaks (for example, displaying the live commentary character 12 in FIG. 2, displaying the spoken text in the speech bubble 14, step S94 in FIG. 16); and selecting a predicted event to be spoken about based on the attributes set for the character.

[0042] According to the eighteenth aspect of the present invention, the computer system can select one predicted event to be the target of speech control, based on the attribute of the character making the speech.

[0043] A 19th invention is a computer system in which, in the above-mentioned computer system, the entertainment is game content played by a player, the speech control means performs the speech control so as to speak to a user, and the player is a person different from the user.

[0044] According to the nineteenth aspect of the present invention, the computer system performs speech control for game content played as a player by a person different from a user who is an observer of entertainment.

[0045] A twentieth invention is the computer system described above, wherein the entertainment is game content played by a player, and the speech control means performs the speech control so as to speak to the player.

[0046] According to the twentieth aspect, the computer system can perform speech control so that speech is directed to the user, the player.

[0047] A 21st invention is a computer system in which the entertainment is game content played by a player, and the player is a computer-controlled player and also the character (for example, the commentator character 12 and player character 18 in Figure 2).

[0048] According to the twenty-first aspect of the present invention, the computer system can execute speech control for making a character, which is a computer-controlled player, speak about the content of his / her game play.

[0049] A 22nd aspect of the invention is the computer system described above, further comprising a storage control means (e.g., the server processing unit 200s, the storage control unit 218 in FIG. 8, the non-realized event data 742 in FIG. 11) for controlling storage of at least a predicted event determined by the determination means to have not occurred among the plurality of predicted events predicted by the prediction means, The speech control means is a computer system that performs the speech control at a given speech timing (for example, the flow from YES in step S50 in FIG. 13 to step S88 in FIG. 14).

[0050] According to the twenty-second aspect of the present invention, the computer system can store predicted events that have been determined by the determining means to have not occurred, and can execute speech control at a given speech timing.

[0051] A twenty-third invention is the computer system, wherein the entertainment has breakpoints in the progression, and the speech control means performs the speech control using the breakpoints as the speech timings.

[0052] According to the twenty-third aspect of the present invention, the computer system can execute speech control at the timing of progress breakpoints in the entertainment.

[0053] A 24th invention is a program for causing a computer system to execute speech control to speak topics related to a given entertainment in text and / or audio based on progress data of the entertainment, the program causing the computer system to function as: a prediction means for predicting a predicted event that may occur in the entertainment at a given prediction timing based on the progress data; a judgment means for determining whether the predicted event has occurred in the entertainment after the prediction timing; and a speech control means for performing the speech control based on the judgment result by the judgment means.

[0054] According to a twenty-fourth aspect of the present invention, it is possible to realize a program that enables a computer system to achieve the same effects as the above-mentioned aspects. [Brief description of the drawings]

[0055] [Figure 1] FIG. 1 is a system configuration diagram showing an example of the configuration of a distribution system. [Diagram 2] FIG. 13 is a diagram showing an example of a screen configuration for a live video. [Diagram 3] 11A and 11B are diagrams for explaining a game operation decision by a player character; [Figure 4] FIG. 4 is a diagram showing a specific example of a predicted event in the situation of FIG. 3. [Diagram 5] FIG. 13 is a diagram for explaining an example of ranking. [Figure 6] FIG. 13 is a diagram for explaining past evaluation comments. [Figure 7] FIG. 13 is a diagram showing an example in which a comment on a reason for retroactive addition is added. [Figure 8] FIG. 2 is a block diagram showing an example of the functional configuration of the server system. [Figure 9] FIG. 4 is a diagram showing examples of programs and data stored in a server storage unit. [Figure 10] FIG. 4 is a diagram showing an example of the data structure of character definition data. [Figure 11] FIG. 4 is a diagram showing an example of the data structure of distribution control data. [Figure 12]FIG. 4 is a functional block diagram showing an example of the functional configuration of a user terminal. [Figure 13] 6 is a flowchart illustrating a flow of a process executed by the server system. [Figure 14] Flowchart continuing from Figure 13. [Figure 15] 11 is a flowchart illustrating the flow of a mention target selection process. [Figure 16] 11 is a flowchart for explaining the flow of a mention target utterance process. [Figure 17] FIG. 11 is a diagram showing an example of a screen configuration of a live video in the second embodiment. [Figure 18] FIG. 11 is a diagram showing an example of the data configuration of game progress control data in the second embodiment. [Figure 19] FIG. 11 is a diagram for explaining prediction in the second embodiment. [Figure 20] FIG. 11 is a diagram for explaining past evaluation comments in the second embodiment. [Figure 21] 10 is a flowchart illustrating a process flow in a server system according to a second embodiment. [Figure 22] FIG. 13 is a diagram for explaining an example of a distribution system and a live video screen configuration in the third embodiment. [Diagram 23] FIG. 13 is a functional block diagram showing an example of the functional configuration of a server system according to a third embodiment. [Figure 24] 13 is a flowchart illustrating a process flow in a server system according to a third embodiment. [Diagram 25] FIG. 13 is a diagram showing an example of a data configuration of prediction model and scene condition data. [Figure 26] FIG. 13 is a functional block diagram showing an example of the functional configuration of a user terminal according to a modified example. [Figure 27] FIG. 4 is a diagram showing an example of the data structure of viewing control data. [Figure 28] 10 is a flowchart illustrating a process flow in a user terminal according to a modified example. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0056] Hereinafter, examples of embodiments of the present invention will be described, but it goes without saying that the forms to which the present invention can be applied are not limited to the following embodiments.

[0057] [First embodiment] FIG. 1 is a system configuration diagram showing an example of the configuration of a distribution system according to this embodiment. The distribution system 1000 is a computer system that generates and distributes video footage of a video game (a type of entertainment content that takes place in a virtual world) with commentary on the progress of the game, that is, so-called game commentary video footage.

[0058] The distribution system 1000 is a computer system including a server system 1100 and user terminals 1500 (1500a, 1500b, ...) connected to each other so as to be capable of data communication via a network 9. The user terminals 1500 serve as a man-machine interface (MMIF).

[0059] The network 9 refers to a communication path that allows data communication. In other words, the network 9 includes a dedicated line (dedicated cable) for direct connection, a LAN (Local Area Network) using Ethernet (registered trademark), etc., as well as a telephone communication network, a cable network, the Internet, etc.

[0060] The server system 1100 performs various controls related to the game commentary videos. That is, the server system 1100 is a computer system that performs various processes such as, for example, control of the progress of the game that is the source of the game commentary videos, control of the generation of commentary by a commentary character, and generation of a game commentary video in which the commentary character appears together with game footage. In addition, the server system 1100 also performs various processes such as control of the management function of the video distribution website, control of the distribution of the game commentary videos, management control of information of registered users, control of the submission acceptance of viewer comments, and control of sharing of viewer comments in the game commentary videos.

[0061] The server system 1100 has a control board 1150 mounted on the main unit. The control board 1150 is equipped with various microprocessors such as a CPU (Central Processing Unit) 1151, a GPU (Graphics Processing Unit), and a DSP (Digital Signal Processor), various IC memories 1152 such as a VRAM, a RAM, and a ROM, and a communication device 1153. Note that some or all of the functions equipped on the control board 1150 may be realized by an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or an SoC (System on a Chip).

[0062] Although the server system 1100 is illustrated as if it were a single server device, it may be realized by a configuration of multiple devices. For example, the server system 1100 may be configured by preparing multiple servers that share functions and connecting them to each other so that they can communicate data via an internal bus or a network 9. The server system 1100 may also include a database or online storage.

[0063] The user terminal 1500 is a computer system for the user 2 to watch the game commentary video, and serves as a man-machine interface in the distribution system 1000.

[0064] The user terminals 1500 (1500a, 1500b, . . . ) are computer systems connectable to the network 9, such as personal computers, smartphones, wearable computers, portable game devices, home game devices, tablet computers, and the like.

[0065] The user terminal 1500 is a computer equipped with an operation input device, an image display device, and a control board 1550 that performs arithmetic processing. Examples of the operation input device include a touch panel 1506, a keyboard, a game controller, a mouse, etc. Examples of the image display device include a touch panel 1506, a head-mounted display, a glass-type display, etc.

[0066] The control board 1550 is equipped with a CPU 1551, various microprocessors such as a GPU and a DSP, various IC memories 1552 such as a VRAM, a RAM and a ROM, a communication module 1553 connected to the network 9, and the like. These elements equipped on the control board 1550 are electrically connected via a bus circuit or the like, and are connected so as to enable reading and writing of data and sending and receiving of signals. A part or all of the control board 1550 may be configured with an ASIC, an FPGA, or an SoC.

[0067] The control board 1550 stores in the IC memory 1552 programs and various data for implementing the functions of the user terminal 1500. The user terminal 1500 executes a predetermined application program to implement the functions of a man-machine interface of the distribution system 1000, and allows the user to watch game commentary videos.

[0068] Specifically, the user terminal 1500 starts up a web browser (application program), accesses a video distribution website managed by the server system 1100, and logs in to view a live video of the game.

[0069] FIG. 2 is a diagram showing an example of a screen configuration of a live video. The live commentary video W2 includes a game image 10, a live commentary character 12, a spoken text display section 14, and a viewer comment display section 16.

[0070] The content of the video game that is the source of the game video 10 is not particularly limited. The example in Fig. 2 shows a baseball game in which computer-controlled players play against each other, but games of other genres may also be used. In other words, the viewing user who watches the live video W2 is a bystander.

[0071] The commentary character 12 is a display body for visualizing the speaker of commentary, commentary, and conversation, and is prepared as a different entity from the characters appearing in the game video 10. The commentary character 12 is selected from among a plurality of candidate characters 13, and speech control and the like are automatically controlled by the server system 1100. The design and number of the commentary characters 12 can be set appropriately. The commentary characters 12 are not limited to the anthropomorphized animal design as shown in FIG. 2, but may be animals themselves, plants, robots, insects, virtual creatures (e.g., dragons, spirits, etc.), geometric symbols, etc. Invisible characters and characters may be omitted.

[0072] The speech text display unit 14 displays, in text form, the contents of the speech uttered by the commentary character 12. The commentary character 12 is displayed facing the user, and the speech text display unit 14 is designed as a speech bubble of the commentary character 12, so that the viewing user sees as if the commentary character 12 is speaking to the user.

[0073] The viewer comment display unit 16 displays the text of the viewer comment posted by the viewer user while watching the game commentary video.

[0074] The commentary character 12 in this embodiment also serves as the player character 18 of the game displayed in the game video 10. In other words, the game commentary video is a live video in which the commentary character 12 plays the game as a computer-controlled player and gives commentary, explanations, and conversations (comments in response to viewers' comments, monologue, etc.).

[0075] 3 is a diagram for explaining the determination of the game operation by the player character 18. The server system 1100 automatically controls the player character 18 as a computer-controlled player in the game of the game video 10. The server system 1100 automatically determines what game operation the player character 18 will perform, and automatically controls the progress of the game.

[0076] In the example of Fig. 3, the game displayed in the game image 10 is a baseball game. One of the two players is a player character 18, and the other is a computer-controlled player controlled by the server system 1100. In other words, the progress of the baseball game is automatically controlled as a game in which two virtually set computer-controlled players play against each other.

[0077] The game progress as shown in Fig. 3 is the bottom of the eighth inning, with two outs, runners on second and third base, the team of the player character 18 on defense with a score of "1", and the opposing team on offense with a score of "0". And right now, the player who is playing in top form today is stepping into the batter's box.

[0078] The server system 1100 automatically determines two game operations: a game operation as the player character 18 and a game operation as the other computer-controlled player. When the game progress corresponds to a given "predicted timing," the two game operations are determined based on a prediction that assumes the game progress at the predicted timing and the past game progress up to the predicted timing.

[0079] The "predicted timing" is the start timing of a unit of game progress based on the game rules, or the arrival timing of a scene that satisfies a certain condition (scene condition) based on the game rules. In the case of a baseball game, the start timing of a unit of game progress would be when a new batter enters the batter's box, the inning changes, etc. The arrival timing of a scene would be when the score is likely to be tied or overtaken, when a squeeze play is likely to occur, etc. Of course, the predicted timing may be set to something other than these. For example, the timing immediately before each pitch to one batter may be set as the predicted timing.

[0080] The "prediction" is performed using, for example, a prediction AI based on machine learning, a search AI such as the αβ method, beam search, or Monte Carlo tree search, a metaheuristic AI such as simulated annealing, a rule-based database, or a decision tree. The server system 1100 performs prediction calculations on multiple predicted events, with each predicted event being treated as a predicted event for one unit of game progress.

[0081] The number of predicted events varies based on the game rules and the game progress. In the example of FIG. 3, three predicted events A, B, and C are shown in detail, but in reality, three or more predicted events are generated.

[0082] FIG. 4 is a diagram showing a specific example of a predicted event in the situation of FIG. One predicted event 40 has a first event element 41 , a second event element 42 , a predicted probability 43 , a predicted point gain / loss 44 , an expected value 45 , an evaluation score 46 , and a ranking 47 .

[0083] The first event element 41 is an assumed game operation indicating a hypothetical game operation selected by the player character 18 at the predicted timing. In the case of a baseball game, for example, the assumed game operation may be a team-level tactical operation based on the playing tendency of the player (player character 18) and depending on the game situation at that time. In the example of FIG. 3, the player character 18 is on the defensive side, so examples of the assumed game operation include pitcher change (with different substitute players), continued pitching, and intentional walk. Also, in terms of pitch units, selection of the type of ball and pitching course may also be assumed game operations. If the player character 18 is on the offensive side, examples of the assumed game operation include hit and run, squeeze, pinch hit, and pinch runner.

[0084] The second event element 42 is a result event that can occur on the assumption that the player character 18 selects the expected game operation of the first event element 41. In the examples of predicted event A (40a) and predicted event B (40b), examples of the second event element 42 include "out (including all situations that result in an out, such as strikeout, fly, grounder, etc.)", "walk", "hit (including first, second, and third base hits)", and "home run". Note that other possible second event elements 42 include a dead ball and a passed ball, but for convenience, these four will be used.

[0085] The second event element 42 may be set in advance as a plurality of candidates according to various game progress situations, and a candidate corresponding to the game progress situation at the prediction timing may be referenced. Alternatively, the second event element 42 may be set as a plurality of candidates according to various game progress situations at the prediction timing, and may be referenced by statistically narrowing down the candidates from a database of past game results of actual baseball games.

[0086] The predicted probability 43 is a value indicating the degree of possibility that a second event element 42 (result event) will occur as a result of a first event element 41 (expected game operation), and is a value predicted / estimated for each combination of the first event element 41 and the second event element 42.

[0087] For example, the probability may be statistically determined based on the results of past actual baseball games. Furthermore, the predicted probability 43 may be further increased or decreased based on the player's play tendency setting data. For example, if the player character 18 is set to be weak against pressure, the predicted probability 43 of the second event element 42 "walk," "hit," and "home run" is increased in a critical situation as shown in FIG. 3. Conversely, if the player character 18 is set to be strong against pressure, the predicted probability 43 of the second event element 42 "out" is increased.

[0088] Furthermore, the predicted probability 43 may be corrected based on the characters present at that time. Specifically, from among separately prepared match performance data of pitchers and batters, the data of match performance corresponding to the pitcher and batter at that time is referred to. If the match performance referred to is favorable to the pitcher, the prediction probability 43 of the second event element 42 "out" is increased. In this case, separately prepared pitcher-specific game situation control data may be referred to, and the increase amount may be adjusted according to the game progress at the time of prediction. For example, for a pitcher whose control rate tends to increase in a critical situation as shown in FIG. 3, the increase amount of the prediction probability 43 of the second event element 42 "out" is made larger than that of a pitcher whose control rate tends to decrease. Conversely, if the match performance referred to is favorable to the batter, the prediction probability 43 of the second event element 42 "walk", "hit", and "home run" may be increased. In this case, separately prepared batting average data by hit type by batter may be referred to, and the increase amount of the second event element 42 "walk", "hit", and "home run" may be adjusted according to the characteristics of the batter.

[0089] In the example of FIG. 4, when comparing predicted event A, predicted event B, and predicted event C, there is a difference in predicted probability 43 due to differences in pitchers even though the combination of first event element 41 and second event element 42 is the same.

[0090] The predicted points gained or lost 44 indicates the points (positive value) or points lost (negative value) that will be gained if the second event element 42 is realized as a result of inputting the anticipated game operation.

[0091] The predicted runs 44 may be calculated based on a database of past game results of actual baseball games, or a past play database that accumulates game results. Specifically, past events may be extracted from these databases using the game progress at the time of the prediction (e.g., combinations of bases with runners on base, batters' batting averages, etc.) as preconditions, and the predicted runs 44 may be derived statistically. Of course, the predicted runs 44 may be determined by a rule-based decision tree that branches on preconditions based on the game progress.

[0092] Additionally, the predicted goals and losses 44 may be adjusted based on the players currently playing. Specifically, the game performance data corresponding to the pitcher and batter at that time is referenced from separately prepared game performance data between pitchers and batters. If the game performance referenced is in the batter's favor, the predicted runs scored 44 for the second event elements 42 "walks," "hits," and "home runs" may be increased. In this case, the amount of increase for each of the second event elements 42 "walks," "hits," and "home runs" may be adjusted to suit the characteristics of the batter by referring to the batting average data for the batter by different hit types separately prepared.

[0093] In the example of FIG. 4, when predicted event A, predicted event B, and predicted event C are compared, there is a difference in predicted runs scored or lost 44 due to differences in pitchers even if the combination of first event element 41 and second event element 42 is the same.

[0094] It is to be noted that only one of the correction of the predicted probability 43 based on the characters at that time and the correction of the predicted points gained or lost 44 based on the characters at that time may be performed.

[0095] The expected value 45 is calculated by multiplying the predicted points gained or lost 44 by the predicted probability 43 for each combination of the first event element 41 and the second event element 42.

[0096] The evaluation score 46 is calculated as the sum of all the expected values ​​45, and represents the overall evaluation of the predicted event for winning the game.

[0097] The order 47 indicates a selection order that is recommended for the player character 18 to proceed with the game more advantageously.

[0098] FIG. 5 is a diagram for explaining an example of ranking. The ranking 47 is calculated by a ranking function f prepared for each character type of the player character 18. The ranking function f includes, as variables, an evaluation value V, a play tendency adaptation correction value h, and a past occurrence count R. Of course, data other than these may also be used as appropriate variables.

[0099] The play tendency adaptation correction value h is determined based on the setting of the character attributes of the player character 18 and the setting of the assumed game operation attributes.

[0100] The character attribute is, for example, the playing tendency of the player character 18 as a player. Specifically, it is set by two-dimensional data of an "aggressive level" indicating the degree of an offensive personality and a "conservative level" indicating the degree of a defensive personality. Of course, further variations may be set. If the game rules allow the player character 18 to be selected from a race other than human (e.g., elf, dwarf, beastman, etc.), the character attribute may be race.

[0101] The expected game operation attributes are set as two-dimensional data of "aggressive level" and "conservative level." The "aggressive level" indicates the degree to which the game operation has an aggressive or aggressive meaning in light of the game rules. The "conservative level" indicates the degree to which the game operation has a passive or defensive meaning in light of the game rules.

[0102] For example, the assumed game operation attributes "aggressive level" and "conservative level" each have five levels, from level "1 (low)" to level "5 (high)." The pitcher change of predicted event A (switch to pitcher A) and predicted event C (switch to pitcher B) may be set to aggressive level = 4 / conservative level = 1. Predicted event B (continuing in the game) may be set to aggressive level = 1 / conservative level = 4.

[0103] The play tendency compatibility correction value h is calculated by character attribute "aggressive level" x expected game operation attribute "aggressive level" + character attribute "conservative level" x expected game operation attribute "conservative level." In other words, the more the character attribute and the expected game operation attribute match, the larger the play tendency compatibility correction value h.

[0104] The past occurrence count R is the number of times that the same combination of the expected game operation (first event element) and the result event (second event element) occurred during game play (in the past). The past occurrence count R may be substituted with the proportion of the same combination among the events that occurred in the past.

[0105] The ranking function f is set to have a higher value as the evaluation value V is larger. The ranking function f is also set to have a higher value as the play tendency adaptation correction value h is larger. The ranking function f is also set to have a higher value as the past occurrence count R is larger. The influence of the past occurrence count R on the ranking corresponds to that of the proverb "Once twice happens, twice happens," and brings a "human feel" to the ranking.

[0106] A different ranking function f is set for each character type of the player character 18, and the play style set for the character is determined by how much weight is placed on the evaluation value V, the play tendency adaptation value h, and the number of past occurrences R in the ranking function f. If weight is placed on the evaluation value V, for example, the play style will be one in which the player thinks things through theoretically. If weight is placed on the play tendency adaptation value h, for example, the play style will be one in which the player prioritizes his / her own preferences and is emotional and intuitive. If weight is placed on the number of past occurrences R, for example, the play style will be one in which the player emphasizes "luck on that day" and "the flow of the game on that day". In other words, the setting of the ranking function f is one element that shapes the play tendency of the computer-controlled player.

[0107] A predicted event 40 with a relatively low ranking can be said to correspond to a tendency-incompatible predicted event that does not match the playing tendency of the player character 18 within the prediction range at that time. A predicted event 40 with a relatively high ranking can be said to correspond to a tendency-compatible predicted event that matches the playing tendency of the player character 18 within the prediction range at that time.

[0108] 3 is a predicted value indicating the degree of possibility that the player character 18 will adopt it as a game operation at that time. For example, it can be calculated by a predetermined function with the rank 47 and the play tendency adaptation correction value h as variables, so that the higher the rank 47, the higher the value becomes, and the higher the play tendency adaptation correction value h, the higher the value becomes.

[0109] When determining the game operation of the player character 18, the server system 1100 adopts the predicted event A (40a) ranked first, and controls the game progress by regarding the assumed game operation "switch to pitcher A" as an operation input by the player character 18. This adopted predicted event 40 is called the "adopted predicted event." The server system 1100 saves history data of the adopted predicted event. The "adopted predicted event" may be selected and determined at random.

[0110] The commentator character 12 is controlled to utter a commentary comment that refers to the adoption predicted event after the prediction timing. In the example of Fig. 3, since the commentator character 12 also serves as the player character 18, the comment that he utters can be a commentary, an explanation, or even a monologue about the thought process that determines the game operation at that time.

[0111] 6 is a diagram for explaining the past evaluation comment. The past evaluation comment is a type of comment uttered by the commentator character 12, and is a comment that evaluates the past progress of the game by looking back on it.

[0112] The game progress state in FIG. 6 is the state after FIG. 3, in which the player character 18 performs a game operation to switch to pitcher A, but the pitcher A hits a home run and the game is overturned.

[0113] The server system 1100 determines whether the adoption predicted event has been realized (occurred) at a given realization confirmation timing. The "realization confirmation timing" is a break in the game progress based on the game rules or a break in a unit of game progress. Specifically, in the case of a baseball game, this corresponds to the occurrence of a hit, an out, a player substitution, a change in inning, a reversal, and the end of the game. Basically, it is the end of the same unit of game progress as the prediction timing, but it does not have to be the same unit. Also, when a viewer posts a comment about a predicted adoption event or a predicted non-adoption event, this may be considered as the realization confirmation timing.

[0114] In the example of Fig. 6, the server system 1100 performs a confirmation of the realization of the fact that the batter who entered the batter's box in Fig. 3 hit a home run, and it is deemed to be the timing for confirmation of the realization of the fact that the realization of the fact that the batter entered the batter's box in Fig. 3 hit a home run. In the example of Fig. 6, the result of the confirmation of the realization of the fact that the realization of the fact that the batter entered the batter's box in Fig. 3 hit a home run is negative.

[0115] If the result of the realization confirmation is negative, the server system 1100 selects a reference target from, for example, the remaining predicted events (predicted event B (40b), predicted event C (40c), ...) predicted at the same prediction timing as the adoption predicted event (predicted event A (40a)) that is the subject of the realization confirmation.

[0116] The requirements for selecting a mention target (requirements for selecting a mention target) can be set appropriately based on the creator's intentions, game rules, etc. For example, any one of the following (1) to (4) can be set. (1) Events predicting non-adoption. (2) Non-adoptable predictive events that were ranked lower than the adoptable predictive events. (3) A non-adopted forecast event that is the same as or similar to an adopted forecast event that satisfies a given performance condition (e.g., 30% or more of past adopted forecast events) among adopted forecast events that occurred before the adopted forecast event that was confirmed to have been realized. (4) A non-adopted forecast event that is the same as or similar to a past adopted forecast event that satisfies a given low-performance condition (e.g., less than 10% of past adopted forecast events) compared to the adopted forecast event that has been confirmed to have been realized.

[0117] The term "unadopted predicted event" as used herein refers to a predicted event (a predicted event other than an adopted predicted event) that was predicted at the same prediction timing as an adopted predicted event that was confirmed to have been realized, and that was not adopted as the origin of a game operation selection at the time it was predicted.

[0118] The server system 1100 selects which of the above mentioned target selection requirements (1) to (4) will be adopted, and selects at least one of the adoption predicted event and the non-adoption predicted event that corresponds to the selected mentioned target selection requirement as a mention target.

[0119] If the above (1) is adopted, the predicted event that was determined not to have been realized (did not occur) is selected. In addition, the adopted predicted event is the predicted event with the highest ranking based on the play tendency compatibility correction value h (see Figure 5), that is, the "tendency compatible predicted event." Therefore, if the above (1) is adopted, and a trend compatible predicted event is included in the predicted events that were determined not to have been realized, the trend compatible predicted event is selected as the predicted event to be mentioned.

[0120] When the above (2) is adopted, the second-ranked predicted event may be selected. Of course, the third-ranked or lower or the lowest-ranked predicted event may also be selected. The ranking of the predicted events is calculated using the play tendency adaptation correction value h based on the character attributes of the player character 18. Therefore, when the above (2) is adopted, the predicted events to be mentioned are selected based on the character attributes set for the player character 18.

[0121] In addition, the play tendency compatibility correction value h is used to determine the ranking, and the ranking of predicted events that do not match the play tendency is calculated to be low. For this reason, it can be said that an unadopted predicted event ranked lower than an adopted predicted event corresponds to a "tendency incompatible predicted event" among the predicted events at that prediction timing. Therefore, when the above (2) is adopted, if a trend incompatible predicted event is included in the predicted events determined to have not been realized, the trend incompatible predicted event is selected as the predicted event to be mentioned. When the above (2) is adopted, the case of selecting the lowest ranked predicted event corresponds exactly to this.

[0122] In addition, if we take note of the fact that the evaluation value V is used to calculate the ranking, when the above (2) is adopted, it can be said that the predicted event to be mentioned is selected based on the degree of likelihood of occurrence (prediction probability 43) of the predicted event that is determined to have not been realized (did not occur).

[0123] When the above (3) and (4) are adopted, the predicted event to be mentioned is selected based on the past playing performance of the player character 18.

[0124] When adopting the mention target selection requirements, "how many to select" and "which to select" may be selected randomly each time. For the above (3) and (4), the settings of the character attributes "aggressive level" and "conservative level" of the player character 18 are referenced. Then, in a character attribute setting where the aggressive level is greater than the conservative level, the above (3) may be selected with priority. In a character attribute setting where the conservative level is greater than the aggressive level, the above (4) may be selected with priority.

[0125] Then, when the server system 1100 has determined the predicted event to be mentioned, it causes the commentary character 12 to utter a comment mentioning the predicted event as a past evaluation comment. In other words, the predicted event selected as the target to be mentioned is the predicted event to be the target of the past evaluation comment, and the realization confirmation timing is also the "utterance timing" of the past evaluation comment. Note that the utterance timing may be set to be a time period appropriately set after the realization confirmation timing. For example, this is suitable when the commentary character 12 is set as a character that reacts slowly to the progress of the game.

[0126] If the above (1) is selected, the past evaluation comment will be a comment that denies the past selection. For example, "(The assumed game operation of the recruitment prediction event) was wrong," or "Was it a mistake to switch to pitcher A?"

[0127] If the above (2) is selected, the past evaluation comment will be a comment to the effect that the corresponding non-approved predicted event should have been selected. For example, "Should I have chosen (the assumed game operation for the non-approved predicted event)?", "Should I have continued pitching?", "Should I have been replaced by pitcher B?", etc.

[0128] If the above (3) is selected, the past evaluation comment will be a comment that the corresponding predicted event should have been selected in light of past performance. For example, "I wonder if I should have done (the predicted game operation mentioned) as usual."

[0129] If the above (4) is selected, the past evaluation comment will be a comment that the corresponding predicted event should have been selected regardless of past performance. For example, "Maybe I should have selected (the predicted game operation being mentioned) once in a while."

[0130] 6 is an example in which predicted event B (40b) is selected as the subject of mention. In the utterance text display unit 14, past evaluation comments that mention predicted event B are displayed together with comments about the fact that the adopted predicted event (predicted event A (40a)) did not come true (did not go well).

[0131] By reading the past evaluation comments evaluating the past game operation choices as text displayed in the spoken text display unit 14 or listening to them as spoken audio, the user feels that the commentator character 12 is more human, which increases interest.

[0132] FIG. 7 is a diagram showing a display example in which a postscriptive reason comment is added to a past evaluation comment. You may add a retroactive reason comment 19 to the past evaluation comment.

[0133] Specifically, you can generate or add a comment19 with a retroactive reason, such as, "Today's pitcher A could only throw fastballs. I predicted an 80% out rate (batting average of 20%), but in today's conditions, his pitch selection would have resulted in at least 6 hits out of 10 batters (name of batter) (40% out rate, 60% batting average). It would have been better to have the previous pitcher continue pitching and get a 60% out rate (40% batting average)."

[0134] The "80% out rate (20% batting average)" portion of the retroactive reason comment 19 may be determined based on the predicted probability 43 of the predicted event A, which is the adopted predicted event.

[0135] The out rate value of "40% out rate (60% batting average)" is a reevaluation value based on the result of allowing a home run. The numerical part of "It would have been better to let him continue pitching with an out rate of 60% (40% batting average)" is the most likely predicted probability of the predicted event B being mentioned.

[0136] The content of the retroactive reason comment 19 is appropriately generated according to the content of the adoption predicted event or the predicted event to be mentioned. For example, not only the predicted probability 43 but also the predicted points scored / lost 44 may be reevaluated, and the expected value 45 and the evaluation value 46 may be corrected based on these values ​​(see FIG. 4). For example, in the example of FIG. 7, the evaluation value 46 of the predicted event A, which is the adoption predicted event, which was "-0.285" in FIG. 4 etc., has been reevaluated to "-1.700". The retroactive reason comment 19 may include content that refers to these reevaluated values. The adoption probability may be left unchanged, or may be reevaluated / corrected like other numerical values.

[0137] The sentence of the retroactive reason comment 19 may be generated by machine learning AI in a format including numerical value insertion parts. Alternatively, a sentence template in a format including numerical value insertion parts prepared in advance by a rule-based decision tree based on the game progress status may be selected. In order to make a retroactive reason comment, a reevaluation value may be intentionally determined.

[0138] The retroactive reason comment 19 includes a retroactive explanation for why the predicted event did not occur, and the prognosticator's own probability value or an equivalent phrase that was assumed at the time of the prediction, giving the impression of a retroactive reason. When a prediction is wrong, people often think about why the prediction was wrong and inadvertently give a retroactive explanation. Therefore, with this configuration, the commentator character's utterances can be felt to be more "human-like."

[0139] Next, the functional configuration will be described. FIG. 8 is a block diagram showing an example of the functional configuration of the server system 1100. As shown in FIG. The server system 1100 includes an operation input unit 100s, a server processing unit 200s, a sound output unit 390s, an image display unit 392s, a communication unit 394s, and a server storage unit 500s.

[0140] The operation input unit 100s is a means for inputting various operations for managing the server system 1100. For example, a keyboard, a touch panel, a mouse, etc., correspond to this.

[0141] The server processing unit 200s is realized by electronic components such as processors serving as arithmetic circuits, such as a CPU, GPU, ASIC, and FPGA, as well as IC memories, and controls input and output of data between each functional unit including the operation input unit 100s and the server storage unit 500s.The server processing unit 200s performs various types of arithmetic processing based on predetermined programs and data, data received from operation input signals from the operation input unit 100s, and the like, and performs integrated control of the server system 1100.

[0142] The server processing unit 200s has a user management unit 202, a video distribution control unit 210, a game progress control unit 212, a prediction unit 214, a determination unit 216, a memory control unit 218, a speech control unit 220, and a live video generation unit 222. It also has a timing unit 280s, a sound generation unit 290s, an image generation unit 292s, and a communication control unit 294s.

[0143] The user management unit 202 performs processing related to user registration procedures for the distribution system 1000 and stores and manages various types of information associated with user accounts.

[0144] The video distribution control unit 210 controls video distribution, and controls the acceptance of comments posted by viewers on videos being distributed, etc. This also includes management control of the video distribution website.

[0145] The game progress control unit 212 has a function as an entertainment progress control unit 213, and performs various controls related to game execution. Specifically, it sets up a game space by arranging background objects in a virtual three-dimensional space, arranges character objects in the game space, and controls the actions of the character characters according to operation inputs by the player. It then judges the game results.

[0146] The prediction unit 214 predicts a predicted event that may occur in the entertainment (a game in this embodiment) at a given prediction timing based on the progress status data of the entertainment.

[0147] The determination unit 216 determines whether or not the predicted event has occurred in the entertainment after the predicted timing.

[0148] The storage control unit 218 performs control to store at least the predicted events that are determined by the determination unit 216 to have not occurred, among the multiple predicted events predicted by the prediction unit 214.

[0149] The speech control unit 220 performs speech control based on the result of the determination by the determination unit 216 .

[0150] The live video generation unit 222 controls the generation of an image including a video of entertainment and a video of a character speaking about the entertainment on one screen. This corresponds to the generation of the live video W2 of a game (see FIG. 2).

[0151] The timekeeping unit 280s uses a system clock to measure various times such as the current date and time and time limits.

[0152] The sound generation unit 290s is realized by executing an IC or software that generates and decodes audio data. The sound generation unit 290s outputs the generated audio signal to the sound output unit 390s. The sound output unit 390s is realized by a speaker or the like, and emits sound based on the audio signal.

[0153] The image generation unit 292s generates images of various management screens for system management of the server system 1100, and outputs image data to the image display unit 392s. The image generation unit 292s also generates a part or all of the images to be displayed on the user terminal 1500. The image display unit 392s is realized by a device that displays images, such as a flat panel display, a head mounted display, or a projector.

[0154] The communication control unit 294s executes data processing related to data communication, and realizes data exchange with an external device via the communication unit 394s. The communication unit 394s realizes communication by connecting to the network 9. For example, it is realized by a wireless communication device, a modem, a TA (terminal adapter), a jack of a wired communication cable, a control circuit, etc. In the example of FIG. 1, the communication device 1153 corresponds to this.

[0155] The server storage unit 500s stores programs and various data for implementing various functions for the server processing unit 200s to comprehensively control the server system 1100. The server storage unit 500s is also used as a working area for the server processing unit 200s, and temporarily stores the results of calculations executed by the server processing unit 200s according to various programs. These functions are implemented by, for example, IC memories such as RAM and ROM, magnetic disks such as hard disks, optical disks such as CD-ROMs and DVDs, online storage, etc. In the example of FIG. 1, these correspond to storage media such as the IC memory 1152 and hard disks mounted on the server system 1100.

[0156] FIG. 9 is a diagram showing an example of programs and data stored in the server storage unit 500s. The server storage unit 500s stores a server program 501, a delivery client program 503, game initial setting data 510, prediction model data 514, predicted event keyword dictionary data 518, and character definition data 520. It also stores game operation attribute data 560, prediction timing determination condition data 562, realization check timing determination condition data 564, user management data 600, delivery control data 700, past occurrence event history data 740, and current date and time 900. The server storage unit 500s also stores other programs and data (for example, timers, counters, various flags, etc.) as appropriate.

[0157] The server program 501 is a program for realizing the functions of the user management unit 202, the video distribution control unit 210, the game progress control unit 212, the prediction unit 214, the determination unit 216, the memory control unit 218, the speech control unit 220, and the live video generation unit 222.

[0158] The distribution client program 503 is an original client program executed by the user terminal 1500 to use the distribution system 1000. When video distribution is provided as a website, a web browsing application can be executed on the user terminal 1500 instead of a dedicated client program, and therefore the distribution client program 503 may be omitted.

[0159] The game initial setting data 510 stores various initial setting data related to the game that is the source of the game video 10 in the game commentary video. For example, character initial setting data 512 for each character in the game (players in a baseball game) is stored.

[0160] One character initial setting data 512 stores various setting information related to the character in association with the character type, for example.

[0161] In the case of a baseball game, the setting information may be ability parameter values ​​such as batting power, defensive power, running ability, ball control, stamina, batting average, favorite type of pitch, etc. Also, statistical setting values ​​based on past performance in the game world (for example, match results by pitcher, bunt success rate, stolen base success rate, earned run average, etc.) may be included in the setting information.

[0162] The prediction model data 514 is basic data for the prediction unit 214 to make a prediction. For example, when making a prediction using a machine-learned prediction AI, the prediction model data 514 is a data group that stores the learning results. When making a prediction using a rule-based prediction AI, the data group that describes a decision tree corresponds to this.

[0163] The prediction model data 514 includes a plurality of models with different prediction capabilities. The prediction unit 214 applies prediction model data 514 that is compatible with the settings of an applied prediction model that is predetermined for the player character 18. Then, prediction is performed based on information on the game progress state at the prediction timing and information on the game progress state (the progress history) up until the prediction timing.

[0164] Therefore, the prediction model data 514 can be said to be computer-controlled player information 515 for the player character 18 who is a computer-controlled player. Also, it can be said to be computer-controlled player play tendency information 516 that brings about the behavior tendency and play tendency of the player character 18.

[0165] The predicted event keyword dictionary data 518 is prepared for each predicted event that can be predicted by the prediction model data 514, and defines a keyword that represents the predicted event. If a viewer comment contains a keyword defined in the dictionary data, it can be determined that the viewer comment refers to the predicted event corresponding to the keyword.

[0166] The character definition data 520 is prepared in advance for each type of commentator character 12 / player character 18 (see FIG. 2), and stores various data related to the corresponding type of character.

[0167] One character definition data 520 includes, for example, a character type 521, player character setting data 530, commentator character setting data 540, and speech sound generation model data 550, as shown in FIG.

[0168] The player character setting data 530 includes a player level 531, a character attribute setting 532, an applied prediction model ID 534, ranking function data 535, and player utterance generation model data 536. Of course, data other than these may be included as appropriate.

[0169] The character attribute setting 532 corresponds to computer-controlled player play tendency information 533 indicating the play tendency of the player character 18, which is a computer-controlled player.

[0170] The player's utterance generation model data 536 is generation model data for uttering comments as the player character 18. For example, in the case of using a generation AI, it is learning model data or rule data. It may be table data in which the generation conditions and the comment text are associated with each other. By setting the learning method and rules, it is possible to make the character utter a comment that reflects its personality. Specifically, even if the content is the same, differences in personality are created by various tendencies such as the way of speaking (choice of words, number of words, etc.), the frequency of spontaneously speaking even when the other person does not speak to them, the degree of talkativeness / silence, the number of filler words and interjections, and the presence or absence of a dialect. Therefore, the player's utterance generation model data 536 can be said to be player's utterance tendency data 537 that determines the tendency of the character to speak as a player.

[0171] Commentator character setting data 540 includes commentator character attribute setting 541 and commentator utterance generation model data 542. Of course, data other than these may also be included as appropriate.

[0172] The commentator character attribute setting 541 stores, as attributes, settings of the selection tendency of the above-mentioned mention target selection requirements (1) to (4). For example, settings of the selection probability by the number of selections of the mention target selection requirements and the selection probability of each of the mention target selection requirements (1) to (4) are stored. The server system 1100 uses these settings when selecting the mention target selection requirements.

[0173] The live commentary speech generation model data 542 is generation model data for uttering comments as the live commentary character 12. For example, in the case of using a generation AI, it is learning model data or rule data. It may be table data in which generation conditions and comment texts correspond to each other. Depending on the learning method and rule settings, it is possible to make the character utter comments that reflect its personality. Therefore, the live commentary speech generation model data 542 can also be said to be live commentary speech tendency data 543 that determines the tendency of the character to speak as a commentator.

[0174] In addition, by using the live commentary speech generation model data 542, it is possible to make a comment that has a conversation in response to the viewer's comment. That is, it is possible to make a conversational comment that follows on the words in the viewer's comment.

[0175] The speech generation model data 550 is generation model data for generating the voice of the character that reads out the spoken text. The tone of voice differs depending on the type of character.

[0176] Returning to FIG. 9, game operation attribute data 560 is prepared for each type of game operation, and includes an aggressive level indicating the degree to which the game operation is an aggressive behavioral choice from the perspective of the game rules, and a conservative level indicating the degree to which the game operation is a conservative behavioral choice.

[0177] The predicted timing determination condition data 562 is data defining the conditions for determining the "predicted timing", and a plurality of data are prepared. When the predicted timing is the start timing of a break in the game progress based on the game rules, the predicted timing determination condition data 562 indicates a condition for determining a break in the game progress, such as a change in batter or inning.

[0178] When the predicted timing is the timing of arrival of a certain scene based on the game rules, the predicted timing determination condition data 562 defines the conditions for recognizing the scene. For example, to define a scene where a run is likely to be scored, the definition may be made by combining the type and number of bases on which runners are located, the batting order and batting average of the batter, the number of hits by the batter in the game, etc. In other words, the predicted timing determination condition data 562 can also be said to be scene condition data 563.

[0179] The realization check timing determination condition data 564 defines conditions for determining the realization check timing. In this embodiment, speech control is performed immediately based on the determination result of the realization check at the realization check timing, so the realization check timing determination condition data 564 can also be said to be speech timing determination condition data 565. In addition, since the realization check timing is regarded as a breakpoint in the progression, the realization check timing determination condition data 564 can also be said to be breakpoint timing determination condition data 566.

[0180] The user management data 600 is prepared for each user who has registered as a viewer in advance, and stores various data associated with the user.

[0181] Various data related to distribution control of game commentary videos is stored in the distribution control data 700. When multiple game commentary videos are distributed at the same time, distribution control data 700 is prepared for each of them.

[0182] FIG. 11 is a diagram showing an example of the data structure of the distribution control data 700. As shown in FIG. The distribution control data 700 includes, for example, a distribution ID 701, a player setting 703, a commentary character type 705, and game progress status data 707. It also includes commentary character control data 709, predicted event data 710, speech history data 730, viewer comment data 732, and commentary video data 734. Of course, data other than these may also be included as appropriate.

[0183] The player setting 703 stores the user account in the case of a user player, and stores the character type 521 of the player character 18 in the case of a computer-controlled player.

[0184] The commentary character type 705 stores the character type 521 of the character to be the commentary character 12 .

[0185] The game progress status data 707 is progress status data for entertainment such as a game, and various data describing the current and past game progress statuses are stored in chronological order in association with the elapsed time from the start of the game. For example, information such as the number of times a game operation selected by each player was performed and the number of times a selection that has a limited number of times set by the game rules (for example, the number of times a pitcher is changed in a baseball game) was performed is accumulated and saved.

[0186] The predicted event data 710 is created for each predicted event and stores various data related to the event. One predicted event data 710 includes, for example, a predicted event ID 711, a predicted date and time 712, predicted detailed data 713, evaluation value data 715, ranking data 716, an adoption flag 717, a predicted event keyword 718, and game progress status information at the time of prediction 719. Of course, data other than these may also be included as appropriate.

[0187] The detailed prediction data 713 is prepared for each combination of the first event element 41 and the second event element 42, and stores the predicted probability 43, predicted points gained or lost 44, and expected value 45 for the combination (see FIG. 4).

[0188] The adoption flag 717 is set when the predicted event is selected as an “adopted predicted event.” The predicted event data 710 with the adoption flag 717 set corresponds to the history data of the adopted predicted event.

[0189] The game progress information 719 at the time of prediction is set appropriately depending on the game content. For example, in the case of a baseball game, it may include information such as score, inning, out count, runner position, strike count, ball count, pitcher, batter, etc. Alternatively, it may be possible to acquire information on the game progress at that time and information on the history of the game progress up to that time from the game progress data 707 based on the elapsed time as the time elapsed since the start of the game.

[0190] The speech history data 730 is created for each utterance of the commentator character 12 or the player character 18, and stores various data related to the utterance. One piece of utterance history data 730 includes, for example, the type of speaking character, the utterance date and time, the utterance text data, and the reference target predicted event ID. The reference target predicted event ID is set when the utterance refers to a predicted event, and indicates the reference target.

[0191] The viewer comment data 732 is created every time a viewer posts a comment. One piece of viewer comment data 732 includes the viewer name, posting date and time, and comment text.

[0192] Returning to Figure 9, the past occurrence event history data 740 is created each time a confirmation is made as to whether or not a predicted adopted event has been realized, and is data that records whether or not the predicted adopted event has been realized, and if not, whether or not a predicted non-adopted event has been realized.

[0193] One piece of past occurrence event history data 740 includes, for example, an adoption predicted event ID, an achievement flag of the adoption predicted event, and an achievement of a non-adoption predicted event ID. The realization flag is changed from "0 (initial value; not realized)" to "1 (realized)" when it is determined that the adoption prediction event that has been confirmed as realized has been realized (occurred). The realized non-adopted predicted event ID indicates a predicted event that has been realized (occurred) among the remaining predicted events predicted at the same prediction timing as the adopted predicted event when the adopted predicted event is not realized.

[0194] The past occurrence event history data 740 is referenced in relation to the calculation of the past occurrence count R, which is one of the variables of the ranking function f. Also, by referring to the past occurrence event history data 740, a part of the predicted event data 710 can be used as non-realized event data 742 that stores predicted events that have been determined not to have occurred (see FIG. 11).

[0195] FIG. 12 is a functional block diagram showing an example of the functional configuration of the user terminal 1500. The user terminal 1500 includes an operation input unit 100 , a device processing unit 200 , a sound output unit 390 , an image display unit 392 , a communication unit 394 , and a terminal storage unit 500 .

[0196] The operation input unit 100 outputs an operation input signal corresponding to various operation inputs made by a user to the device processing unit 200. For example, the operation input unit 100 can be realized by a push switch, a touch panel, a joystick, a touch pad, a track ball, an acceleration sensor, a gyro, or the like.

[0197] The device processing unit 200 is realized by electronic components such as a microprocessor such as a CPU or GPU, or an IC memory, and controls data input / output between each functional unit including the operation input unit 100 and the terminal storage unit 500. Then, it executes various types of arithmetic processing based on predetermined programs and data, operation input signals from the operation input unit 100, and various types of data received from the server system 1100, and controls the operation of the user terminal 1500.

[0198] The terminal processing unit 200 has a viewing control unit 250, a timing unit 280, a sound generation unit 290, an image generation unit 292, and a communication control unit 294.

[0199] The viewing control unit 250 connects to the server system 1100, receives data of the game commentary video, and performs various controls to enable viewing on the terminal. Since this can also be considered as control as a client in the distribution system 1000, the viewing control unit 250 can also be considered as a client control unit 260.

[0200] The client control unit 260 has an operation input information providing unit 261 and a display control unit 262. The operation input information providing unit 261 performs control to transmit operation input information to the server system 1100 in response to an input from the operation input unit 100. The display control unit 262 performs control to display various images based on data received from the server system 1100.

[0201] The timekeeping unit 280 uses a system clock to keep track of the current date and time, time limit, and the like.

[0202] The sound generation unit 290 is realized by, for example, a digital signal processor (DSP), a processor such as a voice synthesis IC, an audio codec capable of reproducing voice files, or the like, and generates sound signals of music, sound effects, and various operation sounds, and outputs them to the sound output unit 390. The sound output unit 390 is realized by a device that outputs (emits sound) based on the sound signal input from the sound generation unit 290, such as a speaker.

[0203] The image generation unit 292 generates and outputs an image signal for displaying an image on the image display unit 392 based on the control of the viewing control unit 250 (client control unit 260). In the example of Fig. 1, this corresponds to a GPU (Graphics Processing Unit), a graphics controller, a graphics board, etc. mounted on the control board 1550. The image display unit 392 is realized by a device for displaying an image, such as a flat panel display, a head mounted display, or a projector.

[0204] The communication control unit 294 executes data processing related to data communication and realizes data exchange with an external device via the communication unit 394. The communication unit 394 realizes communication by connecting to the network 9. For example, this is realized by a wireless communication device, a modem, a TA (terminal adapter), a jack for a wired communication cable, a control circuit, etc. In the example of FIG. 1, this corresponds to the communication module 1553.

[0205] The terminal storage unit 500 stores programs for causing the terminal processing unit 200 to realize given functions, various data, and the like. It is also used as a working area for the terminal processing unit 200, and temporarily stores the results of calculations executed by the terminal processing unit 200 according to various programs, input data input from the operation input unit 100, and the like. These functions are realized by, for example, IC memories such as RAM and ROM, magnetic disks such as hard disks, optical disks such as CD-ROMs and DVDs, and the like. In the example of FIG. 1, this corresponds to the IC memory 1552 mounted on the control board 1550.

[0206] Specifically, the terminal storage unit 500 stores a viewing program 800 (client program 802), received data 810, viewing control data 812, and a current date and time 900. Of course, other data can also be stored as appropriate. The viewing program 800 causes the user terminal 1500 to function as a viewing control unit 250 for the virtual space service.

[0207] 13 and 14 are flowcharts for explaining the flow of processing executed by the server system 1100. FIG.

[0208] 13, the server system 1100 first selects a commentary character 12 from among the candidate characters 13 (step S10), and then selects a player character 18 (step S12). In a baseball game, two player characters 18 are selected, one of which is the commentary character 12. Then, the server system 1100 starts controlling the game progress (step S14).

[0209] Next, the server system 1100 starts generating and distributing the live commentary video W2 (step S16), and starts accepting viewer comments and recording the display and history of the posted viewer comments (step S18). Then, the server system 1100 starts controlling the speech of the live commentary character 12 (step S20). This causes the live commentary character 12 to make comments that reflect the viewer comments.

[0210] When the prediction timing arrives (YES in step S30), the server system 1100 executes a prediction based on the game progress and generates a plurality of predicted event data 710 (step S32). In making the prediction, various parameters describing the game progress at the prediction timing are referenced. Also, various parameters describing the past game progress up to the prediction timing (e.g., how many pitcher substitutions have been made out of the limited number of pitcher substitutions), i.e., the history, may also be referenced.

[0211] The server system 1100 selects one of the predicted events as an adopted predicted event (step S34), and controls the player to make an utterance referring to the adopted predicted event (step S36). Then, the game operation by the player character 18 at the predicted timing is set as the expected game operation of the adopted predicted event (see FIG. 3) (step S38).

[0212] The server system 1100 judges whether it is the timing to check the realization (step S50). Specifically, in the case of a baseball game, occurrence of a hit, occurrence of an out, substitution of a player, change of inning, end of the game, etc. are judged as the timing to check the realization. Also, when a viewer comment including a keyword defined in the predicted event keyword dictionary data 518 is posted, it is judged as the timing to check the realization.

[0213] If it is determined that it is time to confirm the realization (YES in step S50), the server system 1100 executes a realization check to determine whether or not the predicted adoption event has been realized (step S52).

[0214] Turning now to FIG. 14, if the result of the realization determination is negative (NO in step S54), the server system 1100 executes a mention target selection process (step S56).

[0215] FIG. 15 is a flowchart illustrating the flow of the mention target selection process. In the mention target selection process, if the realization confirmation timing in step S52 is due to a viewer comment (YES in step S70), the server system 1100 searches for predicted events mentioned in the viewer comment (step S72).

[0216] Specifically, a predicted event corresponding to the predicted event keyword dictionary data 518 that defines the keyword determined to be included in the viewer comment is searched for among the remaining predicted events predicted at the same prediction timing as the adopted predicted event.

[0217] Then, if there is a predictable event mentioned in the viewer comment (YES in step S74), the server system 1100 selects the retrieved predictable event as the subject of the comment (step S76).

[0218] If the realization confirmation timing is not due to a viewer comment (NO in step S70), the server system 1100 refers to the commentary character attribute setting 541 (see FIG. 10) of the commentary character 12. Then, the commentary character attribute setting 541 is used to determine the number and types of mention target selection requirements (step S80). It is determined "how many to adopt" and "which to adopt" of the four mention target selection requirements.

[0219] Next, the predicted events that satisfy the mentioned target requirements of the determined type are selected as the mentioned target based on their respective rankings from among the remaining predicted events predicted at the same prediction timing as the adopted predicted event (step S82).

[0220] Specifically, by making a priority selection based on the ranking derived by the ranking function f using the play tendency adaptation correction value h (see FIG. 5), a predicted event that satisfies the requirement to be mentioned is selected based on the character attribute set for the player character 18. Similarly, by making a priority selection based on the ranking derived by the ranking function f using the evaluation value V (the predicted probability is used for calculation), a predicted event to be uttered is selected based on the predicted probability (degree of occurrence possibility) of each predicted event. Similarly, by making a priority selection based on the ranking derived by the ranking function f using the past occurrence count R, a predicted event to be uttered is selected based on past occurrence events.

[0221] Returning to FIG. 14, the server system 1100 executes the mention target utterance process (step S88).

[0222] FIG. 16 is a flowchart illustrating the flow of the mention utterance process. The server system 1100 first generates text of a past evaluation comment that mentions the predicted event to be mentioned (step S90), and causes the commentator character 12 to perform a speech action while emitting a speech sound that reads out the past evaluation comment (step S92). Then, the server system 1100 causes the speech text display unit 14 to display the text of the past evaluation comment (step S94).

[0223] Next, the server system 1100 determines whether to add a comment for the reason for adding later (step S100). For example, a random lottery process may be performed to determine whether to add or not. Alternatively, a positive determination may be made when the game progress satisfies a predetermined condition for adding later (for example, in a baseball game, the game is overtaken, a home run is hit, etc.).

[0224] When adding a retroactive reason comment (YES in step S100), the server system 1100 generates text for the retroactive reason comment (step S102), and determines a reevaluated value of the predicted probability of the unrealized adoption predicted event and a numerical value to be inserted into the text for the retroactive reason comment (step S104).

[0225] The server system 1100 then causes the commentator character 12 to make a speech action while emitting a speech voice reading out the added reason comment into which a numerical value such as the re-evaluation value is inserted (step S106). Furthermore, the server system 1100 causes the speech text display unit 14 to display the text of the added reason comment into which a numerical value such as the re-evaluation value is inserted (step S108).

[0226] Returning to FIG. 14, the server system 1100 generates and updates the utterance history data 730 (step S118). The server system 1100 repeats steps S30 to S118 until the game ends (NO in step S120). When the game ends (YES in step S120), the server system 1100 ends the distribution (step S122).

[0227] As described above, according to this embodiment, it is possible to have a character who speaks about the contents of entertainment such as a video game make past evaluation comments.

[0228] The past evaluation comment refers to a past adoption predicted event that was not realized, and refers to the adoption predicted event and other unadopted predicted events. The past evaluation comment corresponds to a person reevaluating a past choice or making a statement that regrets a choice. Therefore, the computer-controlled live commentary character 12 feels more human to the viewer user who sees the comment, and interest is improved. In speech control related to entertainment, speech other than the ongoing current situation is possible, and speech control that improves interest can be realized.

[0229] Second Embodiment Next, a second embodiment to which the present invention is applied will be described. The second embodiment is basically realized in the same manner as the first embodiment, but is different in that the game player is the user 2, and the commentator character 12 does not also serve as the player character 18. Note that the same components as those in the first embodiment are given the same reference numerals as those in the first embodiment, and duplicated explanations are omitted, and hereinafter, differences from the first embodiment will be mainly described.

[0230] FIG. 17 is a diagram showing an example of the screen configuration of a live video W4 in the second embodiment. The game video 10 displayed in the live video W4 is a game screen in which a user 2 is playing as a player 3 (user player) using a user terminal 1500 as an input device for game operation. The game genre is action RPG. The player 3 operates multiple player characters 4 (4a, 4b, 4c) and aims to clear the game while surviving encounters with enemy NPCs 6 (6a, 6B) encountered in the game space by making full use of weapons, magic, and items. The user 2, who is the player 3, can purchase items and use them in the game.

[0231] The commentator character 12 utters comments to give commentary or impressions on the game play, as if speaking to the player 3 who is the user 2, and the text of the utterance is displayed in the utterance text display unit .

[0232] As shown in FIG. 18, in the second embodiment, game progress status data 707, item purchase history data 750, owned item data 752, and game operation history data 754 are linked to the user account of player 3 and stored and managed.

[0233] The game operation history data 754 is one of the player information 755. The game operation history data 754 is created every time the player 3 inputs an operation, and the operation date and time, the game operation type, and the game progress status information at the time of operation are stored in association with each other. The game progress status information at the time of operation indicates the situation in which the game operation type was selected by the player 3. In the case of an action RPG, for example, various status values ​​of the player character 4, the type and number of enemy NPCs 6, the type and number of items possessed, etc. may be used. By performing statistical processing for each type of game operation, the operation tendency of the user player appears. In other words, the game operation history data 754 is the play tendency information 756 of the user player.

[0234] The item purchase history data 750, owned item data 752, and game operation history data 754 are saved as save data in the user management data 600 when the game ends, and are read out and carried over the next time the game is played.

[0235] FIG. 19 is a diagram for explaining prediction in the second embodiment. The server system 1100 of the second embodiment predicts a possible action of the player 3 based on the game progress state at the predicted timing and the past game progress state up to the predicted timing. Also, the server system 1100 predicts a possible action of the player 3 based on the item purchase history data 750, the owned item data 752, and the game operation history data 754.

[0236] In the game progress state of FIG. 19, one of the player characters has only a little HP remaining, and if this continues, the player character will be unable to act. As a result of prediction, it is predicted that the player 3 will perform predicted event D (40d; using a recovery medicine item), predicted event E (40e; using recovery magic), and predicted event F (40f; doing nothing to deal with the fact that the remaining HP is low). The "adoption probability" of each of these predicted events 40 (40d, 40e, 40f) is a predicted value that indicates the degree of possibility that it will be adopted as a game operation at that time based on the play tendency information 756 of the player 3 so far.

[0237] In predicted event D (40d), the expected game operation of the first event element is "use a recovery medicine item", and the result event of the second event element is "recover the set recovery amount of the item". The probability of adoption is "75%". When the expected game operation is executed, the result event will be realized 100%, so the expected value is "set recovery amount of the item" x 0.75. The ranking is 1st place.

[0238] In predicted event E (40e), the expected game operation of the first event element is "use recovery magic", and the result event of the second event element is "recover only the magic effect". The probability of adoption is "24%". When the expected game operation is executed, the result event will be realized 100%, so the expected value is "effect amount" x 0.24. The ranking is second.

[0239] In predicted event F (40f), the expected game operation of the first event element is "do nothing," and the result event of the second event element is "no recovery." The adoption probability is "1%." When the expected game operation is executed, the result event will be realized 100%, so the expected value is "no recovery = 0" x 0.01. The ranking is third.

[0240] The effectiveness of recovery magic may be determined each time through a lottery process. For example, the recovery amount and recovery rate as the effect are determined by a lottery process using variables such as the character level of the player character, magic skill level, and remaining MP. The player level and item purchase history may also be used as variables for the lottery process. Then, depending on the lottery result, stages of effectiveness may be set. In this case, for the predicted event E, stages of effectiveness may be set as second event elements, and the predicted probability for each second event element may be the lottery probability for each stage.

[0241] The server system 1100 regards predicted events whose adoption probability is a predetermined high probability value as tendency-matching predicted events that match the playing tendency based on the playing tendency information 756, and regards the predicted event with the highest adoption probability among the tendency-matching predicted events as the "adopted predicted event."

[0242] Then, the server system 1100 causes the commentary character 12 to make a speech referring to the adoption prediction event.

[0243] FIG. 20 is a diagram for explaining past evaluation comments in the second embodiment. If the result of the determination as to whether the predicted event to be adopted will be realized is negative at the realization confirmation timing, the server system 1100 selects a reference target and causes the live commentary character 12 to utter a past evaluation comment that refers to the predicted event of the reference target.

[0244] In the second embodiment, since the player 3 is the user 2, the mention target selection requirements may be the following (α) to (δ). (α) Prediction of non-adoption events. (β) An unadopted predicted event in which the predicted event is realized. (γ) A non-adopted predicted event that is the same as or similar to an adopted predicted event that satisfies a given performance condition (e.g., 30% or more of past adopted predicted events) among past adopted predicted events that are more likely to be realized than the adopted predicted event that has been confirmed to have been realized. (δ) A non-adopted forecast event that is the same as or similar to a past adopted forecast event that satisfies a given low performance condition (e.g., less than 10% of past adopted forecast events) compared to the adopted forecast event that has been confirmed to have been realized.

[0245] The above (α) is a predicted event that has been determined to have not been realized (did not occur), and corresponds to one predicted event selected based on the game operation history data 754 (player information 755). The predicted event corresponding to the above (γ) or (δ) corresponds to one predicted event selected based on the player information 755 of player 3 from among the predicted events determined to have not been realized (did not occur).

[0246] In the example of FIG. 20, the predicted event E (40e) corresponding to the above (β) is selected as the mention target.

[0247] When the predicted event to be mentioned improves the game performance, the past evaluation comment is generated to have positive content, for example, as in the example of the spoken text display unit 14. Conversely, when the predicted event to be mentioned reduces the game performance of the player 3 (user player), the past evaluation comment is generated to have negative content.

[0248] The functional configuration of the server system 1100 in the second embodiment is similar to that in the first embodiment.

[0249] 21 is a flowchart for explaining the flow of processing in the server system 1100 of the second embodiment. The flow of processing in the second embodiment is basically the same as that in the first embodiment. However, since the player 3 in the second embodiment is the user 2 and a commentary is added to the game play screen, the selection of the player character 18 (step S12) in the first embodiment is omitted. Also, for the same reason, step S38 in the first embodiment is omitted.

[0250] Therefore, according to the second embodiment, the same effects as those of the first embodiment can be obtained.

[0251] Third Embodiment Next, a third embodiment to which the present invention is applied will be described. Note that components similar to those in the first and second embodiments are given the same reference numerals as in the first and second embodiments, and duplicated descriptions will be omitted. Differences from the first and second embodiments will be mainly described.

[0252] 22, in a distribution system 1000C of the third embodiment, a server system 1100C acquires a real video 20 and situation data 22 from an external distribution system 1200. The real video 20 is a video of entertainment in the real world (for example, a baseball game, which is content that takes place in the real world).

[0253] The external distribution system 1200 is a computer system that distributes moving images on demand or live. The situation data 22 is information for displaying the progress of the content or events occurring in the content as text on the subtitle display unit 24 or as numerical values ​​or symbols on the data display unit 26 .

[0254] The server system 1100C generates a live video W6 in which the real video 20 acquired from the external distribution system 1200 is replaced with the game video 10, and distributes the live video W6 to the user terminal 1500. In the live video W6, the live character 12 utters live commentary about the entertainment in the real world occurring in the real video 20.

[0255] 23 is a functional block diagram showing an example of the functional configuration of the server system 1100C of the third embodiment. The server system 1100C basically has the same functional configuration as the server system 1100 of the first embodiment, but the game progress control unit 212 is omitted and the server system 1100C has a real image acquisition control unit 224 and a progress status analysis unit 226.

[0256] The programs and data stored in the server storage unit 500s of the server system 1100C are basically the same as those in the first embodiment (see FIG. 9). The real image acquisition control unit 224 and the progress status analysis unit 226 are realized by reading and executing the server program 501 in the third embodiment in the server processing unit 200s.

[0257] The real image acquisition control unit 224 performs control to acquire real images and information about the images from outside the distribution system 1000. Specifically, it performs control to access the external distribution system 1200 and acquire the real image 20 and situation data 22 distributed and provided by the system. When the external distribution system 1200 manages and distributes a video distribution website, the real image acquisition control unit 224 is realized by using a web browser or the API of the external distribution system 1200.

[0258] The progress analysis unit 226 analyzes the real video 20 to generate entertainment progress data equivalent to the situation data 22. For example, this is realized by a generation AI that performs machine learning using video of a specific sporting event as learning material.

[0259] The progress analysis unit 226 is useful in cases where the real video 20 can be acquired but the situation data 22 is not provided. For example, predetermined entertainment video data (video data of past sports games prepared for on-demand) that can be played back in the distribution system 1000 is prepared, and the progress analysis unit 226 analyzes this. Entertainment progress data equivalent to the situation data 22 may then be prepared later, and a live video W6 may be generated and distributed by playing them together.

[0260] FIG. 24 is a flowchart for explaining the flow of processing in the server system 1100C. Basically, the flow is the same as that of the first embodiment. However, in the second embodiment, real images (images of entertainment taking place in the real world) are used instead of the game images 10. Therefore, the steps related to the game (steps S12 and S14 in the first embodiment) are omitted, and instead the server system 1100C starts acquiring the real images 20 and the situation data 22 (step S15). Also, for the same reason, step S38 in the first embodiment is also omitted.

[0261] According to the third embodiment, the same effects as those of the first embodiment can be obtained.

[0262] [Modifications] Although examples of embodiments to which the present invention is applied have been described above, the forms to which the present invention can be applied are not limited to the above-described forms, and components can be added, omitted, or modified as appropriate.

[0263] (Variation 1) In the first embodiment, an example is shown in which the commentary character 12 and the player character 18 are used together, but a configuration in which the commentary character 12 and the player character 18 are separate characters is also possible.

[0264] (Variation 2) In the first embodiment, all the players of the multiplayer game are illustrated as computer-controlled players, but one or more of the players may be users 2 who are watching or spectating the live video W2. They may also be people (e.g., other users) different from the users who are watching or spectating the live video W2.

[0265] (Variation 3) In the first embodiment, the prediction model data 514 (see FIG. 9) and the prediction timing determination condition data 562 (scene condition data 563) are illustrated separately, but depending on the game content, they may be configured together.

[0266] Specifically, instead of the prediction model data 514 and the prediction timing determination condition data 562, prediction model / scene condition data 770 is used that combines both, as shown in Fig. 25. The prediction model / scene condition data 770 includes a scene condition 772 and a plurality of predicted event candidates 774 (774a, 774b, ...) that will occur thereafter. When making a prediction (step S32; see Fig. 13), the condition data of the scene condition 772 that matches the progress of the entertainment is searched for, and one of the predicted event candidates 774 (774a, 774b, ...) is selected as the predicted event.

[0267] This configuration is suitable for games in which players take turns drawing cards from the tableau and discarding cards from their hands to create a winning hand (e.g., Seven Bridge, Gin Rummy, Koi Koi, Hanafuda, etc.).

[0268] In the scene conditions 772, a given scene may be defined using the number of cards remaining on the table, the type and number of cards exposed to the table, the composition of the hand, etc., and in the predicted event candidates 774, combinations of the types of cards that can be drawn and the hands that can be made with them may be defined as predicted events.

[0269] The scene defined by the scene condition 772, particularly the number of cards remaining on the table and the number of cards exposed to the table, is the result of the progress of the game up to that point.

[0270] It can also be said that the prediction using the prediction model and scene condition data 770 is a prediction of a predicted event based on the progress of the situation up to the satisfaction of the scene conditions.

[0271] This configuration can also be applied to rock-paper-scissors. For example, a scene condition 772 may specify "start of rock-paper-scissors," and a combination of one type of fist (rock / paper / scissors) and another type of fist may be set as a predicted event candidate 774 that occurs after that.

[0272] (Variation 4) For example, in the above embodiment, the speech control of the commentary character 12 is realized by the server system 1100, but the present invention is not limited to this. The speech control of the commentary character 12 may be realized by the user terminal 1500, excluding elements related to the viewer comments, and the commentary video may be generated by the user terminal 1500.

[0273] For example, based on the third embodiment, the user terminal 1500D directly acquires the real image 20 and the situation data 22 from the external distribution system 1200, and the user terminal 1500 controls the speech of the commentary character 12, generates the commentary video W6, and controls the display.

[0274] The functional configuration of the user terminal 1500D in this configuration includes, for example, a prediction unit 214, a judgment unit 216, a memory control unit 218, a speech control unit 220, a live video generation unit 222, a real image acquisition control unit 224, and a progress analysis unit 226, as shown in FIG. 26.

[0275] The terminal storage unit 500 in this configuration stores a live video viewing program 505, prediction model data 514, character definition data 520, and prediction timing determination condition data 562. It also stores realization check timing determination condition data 564, viewing control data 700D, and current date and time 900. Of course, data other than these can also be stored as appropriate.

[0276] The viewing control data 700D corresponds to the distribution control data 700 in the first embodiment, and stores various data related to viewing of live video on the user terminal 1500D.

[0277] 27, the viewing control data 700D includes commentary character type 705, commentary character control data 709, predicted event data 710, speech history data 730, and acquired real video data 820. It also includes acquired situation data 822 (corresponding to situation data 22 in FIG. 22) obtained by progress analysis unit 226, and viewing video data 824. Of course, data other than these may also be included as appropriate.

[0278] 28 is a flowchart for explaining the flow of processing in the user terminal 1500D. Basically, the flow is the same as the flow of processing executed by the server system 1100C in the third embodiment, but differs in that the processing is performed by the user terminal 1500D.

[0279] In addition, in this configuration, live video distribution, and viewer comments are not accepted and shared. Therefore, instead of step S16 in the third embodiment, a live video generation step (step S17) is executed. Also, step S18 in the third embodiment, which is related to the acceptance and sharing of viewer comments, is omitted.

[0280] (Variation 5) In the above embodiment, the evaluation value V, the play tendency adaptation correction value h, and the past occurrence count R are used as variables of the ranking function f, but other data may be used as appropriate. For example, the number of votes by viewers may be included as a variable.

[0281] Specifically, in the prediction process (step S32 in FIG. 13), the commentator character 12 is made to speak the content of the predicted event, and is also made to speak a comment encouraging viewers to vote as a viewer comment on which predicted event is most likely to come true.

[0282] For each viewer comment posted within a specified time from the utterance, the system determines whether the words used in the comment are identical or similar to the words in the content of the predicted event, and counts which predicted event the viewer comment is about. These are tallied by predicted event to obtain the number of votes. Then, the ranking of the predicted event is calculated using a ranking function f that includes the number of votes for each predicted event as a variable. The ranking function f in this configuration may be set so that the higher the number of votes, the higher the ranking. [Explanation of symbols]

[0283] 10. Game footage 12...Commentary character 14...Speech text display unit 18...Player character 200s…Server processing section 212...Game progress control unit 213…Entertainment Progress Control Unit 214…Prediction Department 216...Judgment section 218...Memory control unit 220…Speech control unit 222…Live video production department 500s…Server storage 501...Server program 510... Game initial setting data 512…Character initial setting data 514…Prediction model data 515...Computer-controlled player information 516...Computer-controlled player play tendency information 530...Setting data for player character 532…Character attribute settings 533...Computer-controlled player play tendency information 534…Applied prediction model ID 536…Player utterance generation model data 537…Player speech tendency data 540...Commentary character setting data 541…Commentary character attribute setting 542…Live speech generation model data 543…Live commentary speech tendency data 550…Speech generation model data 560...Game operation attribute data 562…Prediction timing judgment condition data 563...Scene condition data 564…Realization confirmation timing judgment condition data 565…Utterance timing judgment condition data 566...Break timing judgment condition data 700...Distribution control data 705…Commentary character type 707...Game progress data 710…Predicted event data 714…Predicted event data 715…Predicted probability 740...Past incident history data 754…Game operation history data 1000…Distribution system 1100...Server system 1500: User terminal

Claims

1. A computer system for controlling speech to utter a topic related to a given entertainment in text and / or voice based on progress data of the entertainment, comprising: A prediction means for predicting a predicted event that may occur in the entertainment at a given prediction timing based on the progress status data; a determination means for determining whether or not the predicted event has occurred in the entertainment after the predicted timing; a speech control means for controlling the speech based on a result of the determination by the determination means; A computer system comprising:

2. the speech control means controls speech content based on speech tendency data set for a given character, thereby controlling the speech so that the character speaks; 2. The computer system of claim 1.

3. The entertainment is content that takes place in a given virtual world; The progress status data is data related to the progress of the content in the virtual world.

2. The computer system of claim 1.

4. the entertainment is content that takes place in the real world; The progress status data is data related to the progress of the content.

2. The computer system of claim 1.

5. the prediction means determines that the prediction timing has arrived when the progress of the entertainment satisfies a given scene condition, and predicts the predicted event; 2. The computer system of claim 1.

6. There are multiple scene conditions, A plurality of predicted event candidates that are candidates for the predicted event are set for each of the scene conditions; the prediction means predicts the predicted event based on the plurality of predicted event candidates set for the scene condition satisfied by the progress status; 6. The computer system of claim 5.

7. the prediction means predicts the predicted event based on the progress state when the scene condition is satisfied; 6. The computer system of claim 5.

8. the prediction means predicts the predicted event based on a history of the progress state until the scene condition is satisfied; 6. The computer system of claim 5.

9. the scene conditions are conditions regarding scenes involving characters appearing in the entertainment, The prediction means predicts the predicted event based on information about the characters.

6. The computer system of claim 5.

10. the entertainment is game content played by a player; The prediction means predicts the predicted event based on player information of the player.

6. The computer system of claim 5.

11. The prediction means predicts the degree of likelihood of the predicted event occurring.

2. The computer system of claim 1.

12. The prediction means predicts a plurality of the predicted events; The speech control means performs the speech control to utter content related to a predicted event determined by the determination means to have not occurred among the plurality of predicted events predicted by the prediction means.

2. The computer system of claim 1.

13. the entertainment is game content played by a player; The prediction means predicts the predicted event; the speech control means selects at least one predicted event from among the predicted events determined by the determination means to have not occurred based on player information of the player, and performs the speech control to utter content related to the selected predicted event.

13. The computer system of claim 12.

14. The player information includes playing tendency information; the prediction means performs the prediction by including a tendency matching prediction event that matches a playing tendency based on the playing tendency information in the predicted prediction events; When the tendency matching prediction event is included in the prediction events determined by the determination means not to have occurred, the speech control means selects the tendency matching prediction event and performs the speech control.

14. The computer system of claim 13.

15. The player information includes playing tendency information; the prediction means performs the prediction by including a trend incompatible predicted event that is incompatible with a playing trend based on the playing trend information in the predicted events to be predicted, When the predicted events determined not to have occurred by the determination means include the tendency incompatible predicted event, the speech control means selects the tendency incompatible predicted event and performs the speech control.

14. The computer system of claim 13.

16. the speech control means selects a predicted event to be spoken about based on a past occurrence of an event in the entertainment from the predicted events determined by the determination means not to have occurred, and performs the speech control.

13. The computer system of claim 12.

17. The prediction means predicts a degree of occurrence likelihood for each of the predicted events, the speech control means selects a predicted event to be uttered based on the degree of occurrence probability of each of the predicted events determined by the determination means not to have occurred, and performs the speech control.

13. The computer system of claim 12.

18. The speech control means includes: performing said speech control so that a given character speaks; selecting a predicted event to be spoken about based on an attribute set for the character; To carry out 13. The computer system of claim 12.

19. The speech control means performs the speech control so as to speak to a user, The player is a different person from the user. A computer system according to any one of claims 10, 13 to 18.

20. the speech control means performs the speech control so as to speak to the player; A computer system according to any one of claims 10, 13 to 18.

21. The player is a computer-controlled player and the character is 20. The computer system of claim 18.

22. a storage control means for controlling storage of at least a predicted event determined by the determination means to have not occurred among the plurality of predicted events predicted by the prediction means; Further comprising: The speech control means performs the speech control at a given speech timing.

13. The computer system of claim 12.

23. The entertainment has a progression of breaks, the speech control means performs the speech control using the break timing as the speech timing; 23. The computer system of claim 22.

24. A program for causing a computer system to execute speech control for uttering a topic related to a given entertainment in text and / or voice based on progress data of the entertainment, the program comprising: a prediction means for predicting a predicted event that may occur in the entertainment at a given prediction timing based on the progress status data; a determination means for determining whether or not the predicted event has occurred in the entertainment after the predicted timing; a speech control means for controlling the speech based on a result of the determination by the determination means; A program for causing the computer system to function as the

Images