Sound source formation system, sound source formation program, and sound source formation method

Abstract

To solve the problem that selecting music matching voice including a word such as announcement is time consuming work.SOLUTION: A sound source formation system 1 includes a music selection part 30 for selecting music related to information on voice including a word based on information on voice including a word and data in which music is associated with a word. The sound source formation system 1 includes a synthesis part 40. The synthesis part 40 synthesizes voice data related to information on voice including a word and music selected by the music selection part 30 to form a sound source.SELECTED DRAWING: Figure 2

Description

Technical Field

[0001] The present disclosure relates to a sound source formation system, a sound source formation program, and a sound source formation method.

Background Art

[0002] There is known a voicemail device that synthesizes background music (hereinafter referred to as BGM) with an announcement voice (for example, Patent Document 1). The voicemail device disclosed in Patent Document 1 forms an information guidance message by synthesizing a pre-selected BGM with the announcement voice.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] By the way, as described in Patent Document 1, music may be added to voice including words such as an announcement. However, selecting music suitable for voice including words from a music database is a time-consuming task.

Means for Solving the Problems

[0005] (1) A sound source formation system that solves the above problems is a sound source formation system that forms a sound source, comprising: a music selection unit that selects music related to the voice-related information as selected music based on voice-related information including words and relational data relating music and related words; and a synthesis unit that forms a sound source by synthesizing the voice data relating to the voice-related information and the selected music. With this configuration, music related to the voice-related information can be selected based on voice-related information including words and relational data relating music and related words. Therefore, the work of selecting music suitable for the content of the voice-related information can be reduced. This makes it possible to shorten the time required to form a sound source.

[0006] (2) The sound source formation system described in (1) above further comprises a word matching unit that matches the related words in the relational data with the words in the voice-related information, and the music selection unit selects music related to the matching words as the selected music based on the matching words that match the related words through matching by the word matching unit and the relational data. With this configuration, the word matching unit matches the related words in the relational data with the words in the voice-related information. The music selection unit then selects music related to the matching words obtained as a result of the matching from the relational data. In this way, music related to voice-related information can be selected from the relational data. Therefore, the work of selecting music suitable for the content of the voice-related information can be reduced. This reduces the time required to form a sound source.

[0007] (3) The sound source formation system described in (2) above further comprises a keyword selection unit that selects a keyword from the matching word, and the music selection unit selects music related to the keyword based on the keyword and the related data. With this configuration, music can be selected based on the keyword. This makes it easier for the music selection unit to select music that is appropriate for the content of the audio-related information.

[0008] (4) In the sound source formation system described in (3) above, priority data is provided, the priority data includes a plurality of registered words, each of the plurality of registered words is assigned a priority indicating its superiority in being selected as the keyword, and the keyword selection unit selects the keyword from the matching words by referring to the priority data. With this configuration, music can be selected based on the keyword with the highest priority. As a result, the music selection unit can efficiently select music that is suitable for the content of the audio-related information.

[0009] (5) The sound source formation system described in (2) above further comprises a range setting unit that sets a range for matching words in the voice-related information, and the word matching unit matches words in the voice-related information within the range set by the range setting unit. With this configuration, a range for matching words in the voice-related information can be set. As a result, the sound source formation system can shorten the time required to complete word matching.

[0010] (6) In the sound source formation system described in (2) above, the music selection unit selects music related to the matching word and the additional information as the selected music based on the matching word and the additional information and the related data. With this configuration, music can be selected based on the audio-related information and the additional information. Therefore, the music selection unit can select music more efficiently.

[0011] (7) In the sound source formation system described in (6) above, the additional information includes the expected weather at the time of sound source use, the season at the time of sound source use, information about the day of sound source use, the location at the time of sound source use, visual information of people at the location of sound source use, the time of sound source use, the purpose of sound source use, the sound environment of the location of sound source use, the style of the store at which the sound source is used, the philosophy behind sound source use, the facilities of the person or organization using the sound source, personal information of people who can hear the sound source, and at least one of the additional keywords that can be set at will. With this configuration, music can be selected not only based on sound-related information but also based on this information.

[0012] (8) In any one of the sound source formation systems described in (1) to (7) above, the relational data is a table or trained data, the table is data in which related words are associated with each of a plurality of songs, and the trained data is data obtained by learning based on a dataset of songs and related words that have been linked together in advance. With this configuration, music can be selected based on the table or trained data.

[0013] (9) In any one of the sound source formation systems described in (1) to (7) above, the voice-related information includes at least one of announcement text information, announcement voice information, and voice information. With this configuration, music suitable for the voice-related information can be selected for the voice-related information which includes at least one of announcement text information, announcement voice information, and voice information.

[0014] (10) A sound source formation program that solves the above problems is a sound source formation program for causing a computer to form a sound source, and causes the computer to execute: a word matching step which matches the related words in relational data which is related to music and related words with words in audio related information; a music selection step which causes the computer to select music related to the matching words based on the matching words and relational data obtained by the word matching step and select music related to the matching words as selected music; and a synthesis step which causes the computer to form a sound source by synthesizing audio data related to the audio related information and the selected music. With this configuration, the word matching step matches the related words in relational data with words in audio related information. The music selection step then selects music related to the matching words obtained by the matching. In this way, music related to matching words can be selected from audio related information. Therefore, the work of selecting music suitable for the content of the audio related information can be reduced. This makes it possible to shorten the time required to form a sound source.

[0015] (11) A sound source formation method that solves the above problems includes: a word matching step of matching related words in relational data, in which music and related words are associated, with words in audio-related information; a music selection step of selecting music related to the matching words as selected music based on the matching words obtained by the word matching step and the relational data; and a synthesis step of forming a sound source by synthesizing audio data related to the audio-related information and the selected music. With this configuration, the word matching step matches related words in relational data with words in audio-related information. The music selection step then selects music related to the matching words obtained by the matching. In this way, music related to matching words can be selected from audio-related information. Therefore, the work of selecting music suitable for the content of the audio-related information can be reduced. This makes it possible to shorten the time required to form a sound source. [Effects of the Invention]

[0016] The sound source formation system, sound source formation program, and sound source formation method of this disclosure can reduce the time required to form a sound source including music. [Brief explanation of the drawing]

[0017] [Figure 1] This is a schematic diagram illustrating the relationship between the sound source generation system and the server. [Figure 2] This is a block diagram showing the configuration of the sound source formation system in the first embodiment. [Figure 3] This is a diagram of the main table for the first embodiment. [Figure 4] This is a diagram of the first table in the first embodiment. [Figure 5] The second table shows the first embodiment. [Figure 6] This figure shows an example of voice-related information in the first embodiment. [Figure 7] This figure shows the priority data in the first embodiment. [Figure 8] This is a diagram illustrating the music selection unit in the first embodiment. [Figure 9] It is a diagram showing the structure of a sound source in the first embodiment. [Figure 10] It is a diagram showing the structure of a sound source for the first modification example. [Figure 11] It is a diagram showing the structure of a sound source for the second modification example. [Figure 12] It is a diagram of the third table for the fourth modification example. [Figure 13] It is a diagram of the fourth table for the fourth modification example. [Figure 14] It is a diagram showing the range for collating words for the fifth modification example.

Mode for Carrying Out the Invention

[0018] <First Embodiment> Referring to FIGS. 1 to 9, the sound source formation system 1 will be described. The sound source formation system forms a sound source. The sound source is sound data reproduced by a broadcast device or an audio playback device. The sound source includes a transmission matter and music 120. The sound source may include an intro inserted at the start portion of the sound source. The sound source may include an outro inserted at the start portion of the sound source. Examples of the transmission matter include greetings, advertisements, catchphrases, guidance, words for attracting attention, etc.

[0019] An example of the sound source is an announcement broadcast by a broadcast device. In one example of the announcement, it includes an intro, a transmission matter, music 120 used as the BGM for the transmission matter, and an outro. The announcement is a transmission matter for the public or visitors.

[0020] Other examples of sound sources are playback data played back by sound equipment. Examples of sound equipment include CD players and MP3 players. The playback data includes a message and music 120 placed before and after the message. Examples of messages include calls to attract customers, advertisements to inform nearby residents that a sales vehicle is nearby, and warnings to draw attention to roads. An example of a call is "Welcome, welcome." An example of an advertisement is "Roasted sweet potatoes." An example of a warning is "Watch out for falling rocks." Announcements are broadcast, for example, in facilities. Examples of facilities include supermarkets, shopping streets, specialty stores, department stores, schools, hospitals, train stations, and airports.

[0021] The sound source formation system 1 forms a sound source based on speech-related information 50. The speech-related information 50 includes announcement text information 50A, announcement voice information, and voice information. The speech-related information 50 may also be information expressed in languages ​​other than Japanese. Examples of languages ​​other than Japanese include Chinese, Korean, and English.

[0022] The audio-related information 50 includes information to be conveyed in order to form a sound source. When an announcement is formed as a sound source, the audio-related information 50 is the information to be conveyed in the announcement. For example, in the case of a department store announcement, the information to be conveyed is a message to customers such as, "Thank you for visiting our store. Today, we are holding an end-of-year sale in the women's clothing department." In this example, the sound source formation system 1 forms an announcement as a sound source based on such a message as the audio-related information 50.

[0023] The audio-related information 50 may be text data or audio data. The announcement text information 50A is text data that represents the information contained in the announcement in text. The announcement audio information is audio data that represents the information contained in the announcement in audio. A specific example of announcement audio information is MP3 data with a recorded human voice. The audio information is audio data that represents information other than the announcement in audio.

[0024] As shown in Figure 1, the sound source formation system 1 connects to the server 3 via the network 2. The server 3 stores multiple music tracks 120 (see Figure 8) used for forming sound sources. Examples of networks include the internet, local networks, and telephone lines.

[0025] As shown in Figure 2, the sound source formation system 1 comprises a word matching unit 10, a keyword selection unit 20, a music selection unit 30, and a synthesis unit 40. The sound source formation system 1 may further include an input unit 4 and an output unit 5.

[0026] The sound source formation system 1 may have pre-existing speech-related information 50 for sound source formation. In one example, the sound source formation system 1 has a library of speech-related information 50. In this case, the sound source formation system 1 includes a storage unit for storing the library of speech-related information 50. The sound source formation system 1 may also acquire speech-related information 50 from an external source. In this case, the sound source formation system 1 includes an input unit 4.

[0027] The input unit 4 acquires audio-related information 50 for creating a sound source. The input unit 4 is configured to connect to a storage unit that stores the audio-related information 50, and acquires the audio-related information 50 from the storage unit. Examples of the storage unit include a hard disk, semiconductor memory, and a server 3 connected to the internet. The storage unit may be located outside the sound source formation system 1 and configured as a cloud-based device, or it may be located within the sound source formation system 1. The output unit 5 outputs the sound source formed by the synthesis unit 40 to the outside. Examples of the output unit 5 include a display device and a sound playback device.

[0028] <Relational Data> The relational data 60 is data used in the music selection unit 30. The relational data 60 is data that associates music 120 with related words 125. In the relational data 60, the music 120 associated with related words 125 includes music such as background music. In the relational data 60, the music 120 associated with related words 125 may also include natural sounds, sound effects such as chimes, etc. The relational data 60 may be data that associates the music data of music 120 with related words 125. The relational data 60 may be data that associates the unique number of music 120 with a word. In the latter case, the unique number and the music 120 corresponding to the unique number are associated. The music 120 is managed by a server 3 separate from the sound source formation system 1. In this embodiment, the relational data 60 is data that associates the unique number of music 120 with related words 125.

[0029] In this embodiment, the relational data 60 is a table 130. The table 130 contains related words 125. Related words 125 are words associated with music 120. The table 130 is data in which music 120 is linked to each of the related words 125. In one example, in the table 130, the unique number of music 120 associated with the related word 125 is linked to the related word 125.

[0030] For example, related words 125 may consist of words related to the facility name 111. Also, related words 125 may consist of words related to the genre. Furthermore, related words 125 may consist of words associated with a specific word.

[0031] Related words 125 indicate the characteristics of music 120, its uses, the seasons in which it is used, etc. Characteristics of music 120 include information such as tempo, genre, and the image it evokes in people. Uses of music 120 include information such as events. Examples of events include sales and sports days.

[0032] Refer to Figures 3 to 5 to explain an example of Table 130. Table 130 includes the main table 130A. Furthermore, table 130 may also include the first table 131 and the second table 132.

[0033] Refer to Figure 3 to explain the main table 130A. As shown in Figure 3, the main table 130A shows the relationship between pre-selected related words 125 and music 120. The related words 125 are pre-selected from a wide variety of communication items. For example, words that characterize the content of the communication item are selected as related words 125.

[0034] Refer to Figure 4 to explain Table 131. As shown in Figure 4, Table 131 shows the relationship between facility name 111 and music 120. Facility name 111 is included in related word 125. In the example of Table 131 shown in Figure 4, wedding venues are associated with music 1 31 and music 4 34. Specialty stores are associated with music 1 31, music 2 32 and music 33. Supermarkets are associated with music 2 32, music 33 and music 4 34. That is, music 1 31 is associated with wedding venues and specialty stores. Music 2 32 is associated with specialty stores and supermarkets. Music 33 is associated with specialty stores and supermarkets. Music 4 34 is associated with wedding venues and supermarkets. According to Table 131, music 120 can be selected by specifying the facility that will use music 120.

[0035] Refer to Figure 5 to explain Table 132. As shown in Figure 5, the second table 132 shows the relationship between related words 125 and the unique number of the music 120 associated with related words 125. In this example, the facility name 111 is included in related words 125. Related words 125 in the second table 132 mainly contain information that describes the characteristics of the music 120.

[0036] The sound source creator or the operator of the sound source creation system 1 registers information about music 120 in the sound source creation system 1 in advance. The sound source creator can register information about music 120 from the registration screen. In one example, the unique number of music 120 is stored in the memory unit of the sound source creation system 1. The music data associated with the unique number is stored in the server 3, linked to the unique number of the music data.

[0037] <Word matching section> As shown in Figure 6, the word matching unit 10 matches the related word 125 with the word in the audio related information 50. In this embodiment, the word matching unit 10 matches each related word 125 in table 130 with the word in the audio related information 50 one by one. For example, the word matching unit 10 matches each related word 125 in main table 130A with the word in the audio related information 50 one by one. A word that matches the related word 125 through matching by the word matching unit 10 is called a "matching word 6".

[0038] Refer to Figure 6 to explain the matching word 6. Figure 6 shows an example of announcement text information 50A, along with the matching word 6 found by the word matching unit 10. The matching word 6 is shown as "6" in Figure 6. In this example, the matching words 6 are: information, restaurant, food court, cafe, points, August, limited time offer, special deal, and specialty store.

[0039] If the audio-related information 50 is audio data, the word matching unit 10 converts the audio-related information 50 into text information and then starts matching the related words 125 with the words in the converted audio-related information 50 based on the table 130.

[0040] <Keyword Selection Section> The keyword selection unit 20 selects keyword 21 from the matching words 6. In this case, the keyword selection unit 20 may select keyword 21 from the matching words 6 by referring to priority data 22. Priority data 22 is reference data for setting the priority of the matching words 6. The keyword selection unit 20 may select all of the matching words 6 as keyword 21, or it may select some of the matching words 6. Furthermore, the keyword selection unit 20 does not have to select keyword 21 from the matching words 6.

[0041] Figure 7 shows an example of priority data 22. Priority data 22 includes multiple registered words 110. In priority data 22, each of the multiple registered words 110 is assigned a priority that indicates its superiority as a keyword 21. The priority is expressed numerically, and the keyword selection unit 20 determines the priority based on that numerical value.

[0042] The registered words 110 in priority data 22 are pre-registered in a memory unit referenced by the sound source formation system 1. The memory unit may be located outside the sound source formation system 1 and configured as a cloud-based device, or it may be located within the sound source formation system 1. The registered words 110 are arbitrarily selected from a dictionary or words used in everyday life. Specifically, words that frequently appear in announcements are registered as registered words 110.

[0043] Priority is set for each registered word 110. Priority is defined as the advantage of being selected as a keyword 21, as described above. A keyword 21 that has an "advantage of being selected as a keyword 21" is a word that is highly related to music 120. If a word that is highly related to music 120 is included in the content of the information conveyed in the audio-related information 50, there is a high probability that music 120 that substantially matches the audio-related information 50 will be selected. For example, in the example in Figure 6, words such as customer, electronic money, and dedicated card are not words that evoke music 120 with the specified characteristics, so these words are words that are not highly related to music 120. In contrast, words such as restaurant, cafe, and August evoke music 120 with the specified characteristics. For example, a restaurant evokes cheerful music. A cafe evokes calm music. August evokes summer music. For this reason, these words are highly related to music 120.

[0044] In the example shown in Figure 6, keyword 21 is indicated as "21". In this example, the keyword selection unit 20 sets the keywords 21 as: restaurant, cafe, points, August, limited time offer, and bargain.

[0045] <Music Selection Department> The music selection unit 30 will be described with reference to Figures 8 and 9.

[0046] As shown in Figure 8, the music selection unit 30 selects a music track 121 from multiple music tracks 120 via the network 2. Specifically, based on the matching word 6 and the relational data 60, it selects a music track 120 associated with the matching word 6 as the music track 121. As mentioned above, the relational data 60 is data that associates music tracks 120 with words. The music selection unit 30 may select one music track 121, select multiple music tracks 121, or select no music track 121 at all.

[0047] If the music selection unit 30 selects multiple music tracks 121, the sound source creator can listen to each of the music tracks 121. The sound source creator may select music track 121A from among the selected music tracks 121 to be combined with the audio data 122 (see Figure 9). Alternatively, the sound source creator does not have to select music track 121A to be combined with the audio data 122. If no music track 121 is selected from the music 120, the sound source creator may create a sound source by combining the audio data 122 and the chime 123 (described later), excluding music track 121.

[0048] The matching word 6 is information for selecting music 120 related to the audio-related information 50 from the relational data 60. Since the matching word 6 is a word included in the audio-related information 50, the music selection unit 30 can select music 120 related to the audio-related information 50 using the matching word 6 and the relational data 60.

[0049] Match word 6 includes a variety of words. Match word 6 may also include words that would be considered noise when selecting music 120 related to the content of the information conveyed in audio-related information 50.

[0050] As described above, the keyword 21 selected from the matching words 6 is a word that is highly related to music 120. By using keyword 21, it is possible to select music 120 that is suitable for audio-related information 50.

[0051] When multiple keywords 21 are selected from audio-related information 50, the multiple music pieces 120 selected using these keywords 21 may not be grouped into a single category. For example, if the audio-related information 50 includes the keyword 21 "Christmas," the audio-related information 50 may simultaneously select classical music, rock music, popular songs, and folk music in relation to Christmas. Therefore, in order to narrow down the number of music pieces 120 to be selected, or to select music pieces 120 that are suitable for the audio-related information 50, additional information may be used in addition to the keywords 21.

[0052] The use of additional information allows for narrowing down the number of music tracks 120 suitable for the audio-related information 50. For example, the music selection unit 30 selects music tracks 121 related to the keyword 21 and the additional information, based on the keyword 21, the additional information, and the related data 60.

[0053] In this embodiment, the music selection unit 30 may, in selecting music 120, refer to additional information in addition to the keyword 21 from the matching word 6. Specifically, the music selection unit 30 selects music 120 associated with the keyword 21 and the additional information as selected music 121 based on the keyword 21, the additional information, and the relational data 60. The keyword 21 and the additional information are information for selecting music 120 from the relational data 60.

[0054] <Additional Information> The additional information includes the expected weather at the time of use, the season at the time of use, information about the day of use, the location at the time of use, visual information about people at the location, the time of use, the purpose of use, the sound environment at the location, the style of the store at which the sound is used, the philosophy behind the use of the sound, the facilities of the person or organization using the sound, personal information of people who may hear the sound, and at least one optional additional keyword. Hereinafter, these items will be referred to as additional information items.

[0055] Methods for registering additional information include registering it by inputting it into the music selection unit 30 via the keyboard through the input screen when searching the sound source formation system 1, registering it by inputting it into the registration screen in advance before searching the sound source formation system 1, and registering it in real time via an external API (Application Programming Interface). The method for registering additional information in this embodiment is to register it by inputting it into the music selection unit 30 via the keyboard through the input screen. The input screen displays items of additional information. The additional information includes the items of additional information along with information related to those items. The additional information is input into the music selection unit 30. An example of additional information includes "predicted weather when using the sound source" as an item and "sunny" as information related to the item.

[0056] The predicted weather when using the sound source indicates the expected weather at the time the sound source is used. For example, if the sound source is used during the rainy season, the predicted weather will be rain. The season in which the sound source is used indicates the season in which the sound source will be used. For example, if the sound source user plans to use the sound source in the spring, "spring" will be entered into the music selection unit 30 as additional information.

[0057] Examples of information regarding the date on which the sound source is used include anniversaries, holidays, and dates of past disasters. The location where the sound source is used indicates the place where the sound source will be used. For example, if the sound source user plans to use the sound source at a train station, the station will be entered into the music selection unit 30 as additional information.

[0058] The visual information of people present at the location where the sound source is used represents the visual information of people present at the location where the sound source is used. This visual information of people includes information on the number of people, generation, attributes, and trends. This visual information of people may also be information assumed to be present at the location where the sound source is used. This visual information of people may be set based on photographic information captured by cameras installed at the location where the sound source is used.

[0059] The people information indicates the number of people present at the location where the sound source is used. If the sound source is created in parallel with its use, people information counted based on photographic information may be input to the music selection unit 30. For example, people information may be counted by a camera installed in a store. In this case, the people information is used as additional information for selecting music 120, allowing the sound source to be changed based on the number of customers.

[0060] Generational information indicates the generational information of people present at the location where the sound source is used. If the sound source is created in parallel with its use, generational information determined based on the filming information may be input to the music selection unit 30. For example, the age range of people captured by cameras installed in a supermarket may be analyzed and used for marketing purposes. In this case, generational information is used as additional information for selecting music 120, allowing the sound source to be changed based on the generation of the people visiting the store.

[0061] Attribute information refers to the attribute information of people present at the location where the sound source is used. Person attribute information refers to the unique characteristics of people present at the location where the sound source is used. For example, a person wearing a business suit is a businessman.

[0062] Trend information indicates the trends among people in the location where the sound source is used. Trend information is information that shows which trends are influencing people in the location where the sound source is used. Trend information can be determined, for example, by referring to current news, and by looking at a person's clothing, the devices they carry, the accessories they wear, etc.

[0063] The time indicated for the use of the sound source shows the time when the sound source is scheduled to be used. Examples of purposes for using sound sources include, for instance, raising awareness, advertising, and making announcements. The sound environment of a location where a sound source is used indicates how easily the sound is transmitted to people in that location. For example, if the sound source is used at a construction site, the noise level is high, making it difficult for announcements to be heard, thus the sound environment is poor. If the sound source is used at an office, the low noise level makes announcements easier to hear, thus the sound environment is good.

[0064] Examples of store styles that use sound sources include, for example, stores catering to young people, formal wear, food and beverage establishments, retail stores, stand-up eateries, and unmanned stores. The philosophy behind the use of sound sources represents the philosophy held by the sound source user regarding its use. Examples of such philosophy include: a feeling of energy, a refreshing feeling, a calming feeling, etc.

[0065] Examples of facilities where individuals or organizations may use the sound sources include supermarkets, offices, factories, hospitals, schools, etc. Examples of personal information of people who may listen to the audio source include gender, age, occupation, affiliation, and other attributes.

[0066] <Operation of the music selection section> As described above, the music selection unit 30 selects music 120 related to the matching word 6 as selected music 121 based on the matching word 6 and the related data 60.

[0067] In another example, the music selection unit 30 selects music 120 related to keyword 21 as selected music 121 based on keyword 21 and related data 60.

[0068] The music selection unit 30 may select music 120 related to the keyword 21 and additional information as selected music 121 based on the keyword 21 and additional information and the related data 60.

[0069] Keyword 21 is a word selected from the matching words 6 that matched based on matching in Table 130. Additional information is information referenced along with Keyword 21 when selecting music 120 from relational data 60. Additional information is information set by the sound source creator.

[0070] As an example of the operation of the music selection unit 30, the case where keyword 21 and additional information are used will be explained. The following example is one in which "point" is selected as keyword 21 from among the matching words 6.

[0071] In the first step, if the location where the sound source is used is set as additional information, the music selection unit 30 refers to the first table 131 or the second table 132. In this embodiment, the first table 131 is referred to. For example, if the location where the sound source is used is a "wedding venue", the music selection unit 30 refers to the first table 131 or the second table 132 and selects music 120 associated with the wedding venue. In the case of the first table 131 shown in Figure 4, the first music 31 and the fourth music 34 are selected.

[0072] In the second step, the music selection unit 30 refers to the second table 132 and selects the music 120 associated with keyword 21. For example, if keyword 21 is "point", the music selection unit 30 selects the music 120 associated with "point". In the case of the second table 132 shown in Figure 5, the first music 31 and the third music 33 are selected.

[0073] The music selection unit 30 selects the music 120 selected in both the first and second steps as candidate music 121. In this example, the music selection unit 30 selects the first music 31, the third music 33, and the fourth music 34 as candidate music 121. The sound source creator then adopts one of the selected music 121 as music 121A to be synthesized with the audio data 122. In this example, the first music 31 is adopted. The music selection unit 30 outputs the final candidate music 121A to the synthesis unit 40. The music selection unit 30 may further output the final candidate music 121A to the output unit 5.

[0074] The above method of selecting music 120 is just one example. The music selection unit 30 may select only music 120 that contains all of the keyword 21 and additional information, or it may select music 120 that contains at least one of the keyword 21 and additional information.

[0075] <Synthesis section> The synthesis unit 40 synthesizes the audio data 122 related to the audio-related information 50 and the selected music 121. The synthesis unit 40 may also synthesize the audio data 122 related to the audio-related information 50, the selected music 121, and the chime 123. The chime 123 is pre-set by the sound source creator. The chime 123 may be selected based on the audio-related information 50 by the functions of the word matching unit 10, the keyword selection unit 20, and the music selection unit 30 described above.

[0076] The audio data 122 related to the audio-related information 50 may be pre-prepared data, or it may be data formed by speech synthesis software from text information (for example, announcement text information 50A).

[0077] As shown in Figure 9, the sound source comprises an intro section, a middle section, and an outro section. Chimes 123 are inserted in the intro and outro sections. Audio data 122 related to audio-related information 50 and selected music 121A adopted by the sound source creator are inserted in the middle section. The selected music 121A is superimposed on the audio data 122 as background music. In one example, the data duration of the selected music 121A is longer than the data duration of the audio data 122. Alternatively, the data duration of the selected music 121A may be shorter than the data duration of the audio data 122. In this case, the selected music 121A may be synthesized by the synthesis unit 40 so that it is played repeatedly for a longer duration than the data duration of the audio data 122.

[0078] The synthesis unit 40 adjusts the audio data 122, chime 123, selected music 121A, and sound source. The adjustments include volume adjustment, noise reduction, changes in musical style, changes in tempo, changes in data duration, and silence removal. The adjustments may be performed automatically or manually.

[0079] Volume adjustments include adjusting the average loudness, adjusting the true peak, and adjusting the gain.

[0080] Noise reduction is the process of removing noise from the audio data 122, chime 123, selected music 121A, and sound source. Noise includes lip noise, white noise, and pop noise.

[0081] The melody, tempo, and data duration can be changed by editing the data for chime 123, selected music 121A, and the sound source.

[0082] The silence cut process is a process that cuts out the silent portions of the data in the audio data 122, chime 123, selected music 121A, and sound source.

[0083] The adjustment may be performed before the synthesis of the voice data 122, chime 123, and selected music 121A. The adjustment may also be performed during the synthesis of the voice data 122, chime 123, and selected music 121A.

[0084] The sound source may be configured as follows: The beginning of the sound source is configured to start with a chime 123 and then fade in the selected music 121A. The middle part of the sound source is configured so that the selected music 121A plays for a while, and then its volume gradually decreases. The end of the middle part of the sound source is configured so that the volume of the selected music 121A is reduced to a predetermined level, then maintained, and after the playback of the audio data 122 ends, the selected music 121A gradually returns to its original level. The end of the sound source is configured so that the selected music 121A plays for a while, then fades out, and then the chime 123 sounds.

[0085] The output unit 5 (see Figure 2) outputs the sound source synthesized by the synthesis unit 40. Possible output file formats include AIFF, AAC, FLAC, WAV, and MP3. In this embodiment, the sound source file format is WAV.

[0086] <Sound Source Creation Program> A sound source generation program causes a computer to generate sound sources. The sound source creation program includes a word matching step, a music selection step, and a synthesis step.

[0087] The word matching step causes the computer to match the related word 125 with the word in the audio-related information 50. In one example, the audio-related information 50 is information acquired by the input unit 4.

[0088] The music selection step causes the computer to select music 120 as selected music 121, based on the matched word 6 and related data 60 that were matched by the matching in the word matching step.

[0089] The synthesis step involves having the computer synthesize the selected music 121 with the audio data 122 related to the audio-related information 50. If multiple selected music 121s are selected in the music selection step, the sound source creator may choose the selected music 121A to be synthesized. The synthesized sound source is output by the output unit 5.

[0090] <Sound source formation method> The sound source formation method is a method for forming a sound source. The sound source formation method includes a word matching step, a music selection step, and a synthesis step.

[0091] The word matching process compares the related words 125 with the words in the audio-related information 50.

[0092] The music selection process selects music 120 as selected music 121 based on the matching word 6 and related data 60 that were matched by the matching process in the word matching process, and which is associated with the matching word 6.

[0093] The synthesis process involves synthesizing the selected music 121 with the audio data 122 related to the audio-related information 50. If multiple selected music 121s are selected in the music selection process, the sound source creator may choose the selected music 121A to be synthesized.

[0094] The operation of this embodiment will now be explained. Sometimes, music is added to audio that includes words, such as announcements. For example, background music (BGM) is often added to announcements in stores. Conversely, sometimes music is intentionally omitted from audio that includes words, such as announcements. For example, there are cases where background music is not necessary for announcements played on trains. Audio creators need to select music that is appropriate for the content of the announcement. They also need to decide whether or not to add music to the announcement based on its content. However, audio creators do not always understand the situations in which the announcements are used. Audio creators need to confirm with the announcement provider how the announcements are used, or infer the situation from the content of the announcement. Because of the need to confirm these things, audio creators spend a considerable amount of time selecting music that is appropriate for the announcement. Even when a music database is available, it is a time-consuming task for audio creators to select music that suits the announcement from the database.

[0095] In this regard, the music selection unit 30 can select music 120 related to the voice-related information 50, such as announcements, as the selected music 121, based on the voice-related information 50 and the related data 60. If there is music 120 related to the voice-related information 50, the music selection unit 30 can select that music 120 as the selected music 121. The music selection unit 30 can also choose not to select music 121 if there is no music 120 related to the voice-related information 50. Therefore, if the sound source formation system 1 outputs the selected music 121, the sound source formation operator can select appropriate music 120 related to the voice-related information 50. If the sound source formation system 1 does not output the selected music 121, the sound source formation operator can understand that no appropriate music 120 related to the voice-related information 50 was found, or that there is no need to add music to the voice-related information 50. Thus, the music selection unit 30 can not only select music suitable for the voice-related information 50, but also provide information on the existence or non-existence of music suitable for the voice-related information 50. The sound source creator can obtain information on the existence of selected music and music suitable for the audio-related information 50. Based on this information, it becomes easier to select music to accompany the audio-related information 50 or to decide not to accompany it. As a result, the total time required for sound source creation can be shortened. The synthesis unit 40 then automatically synthesizes the selected music 121 selected by the music selection unit 30 with the audio data 122. This further shortens the time required to create the sound source.

[0096] The effects of this embodiment will now be explained. (1) The sound source formation system 1 includes a music selection unit 30 that selects music 120 related to the voice-related information 50 as selected music 121 based on voice-related information 50 including words and relational data 60 relating music 120 and related words 125. The sound source formation system 1 also includes a synthesis unit 40 that forms a sound source by synthesizing voice data 122 related to the voice-related information 50 and the selected music 121. With this configuration, music 120 related to the voice-related information 50 can be selected based on voice-related information 50 including words and relational data 60 relating music 120 and related words 125. This reduces the work of selecting music 120 that is suitable for the content of the voice-related information 50. This shortens the time required to form a sound source.

[0097] (2) The sound source formation system 1 further includes a word matching unit 10 that matches the related word 125 in the relation data 60 with the word in the voice-related information 50. The music selection unit 30 selects music 120 related to the matching word 6 as selected music 121 based on the matching word 6 that matches the related word 125 through matching by the word matching unit 10 and the relation data 60. With this configuration, the word matching unit 10 matches the related word 125 in the relation data 60 with the word in the voice-related information 50. Then, the music selection unit 30 selects music 120 related to the matching word 6 obtained as a result of the matching from the relation data 60. In this way, music 120 related to the voice-related information 50 can be selected from the relation data 60. Therefore, the work of selecting music 120 suitable for the content of the voice-related information 50 can be reduced. This reduces the time required to form a sound source.

[0098] (3) The sound source formation system 1 further includes a keyword selection unit 20 that selects a keyword 21 from the matching words 6. The music selection unit 30 selects music 120 related to the keyword 21 based on the keyword 21 and the related data 60. With this configuration, music 120 can be selected based on the keyword 21. This makes it easier for the music selection unit 30 to select music 120 that is suitable for the content of the audio related information 50.

[0099] (4) The sound source formation system 1 has priority data 22. The priority data 22 includes a plurality of registered words 110. Each of the plurality of registered words 110 is assigned a priority that indicates its superiority in being selected as a keyword 21. The keyword selection unit 20 selects a keyword 21 from the matching words 6 by referring to the priority data 22. With this configuration, music 120 can be selected based on the keyword 21 with the highest priority. Therefore, the music selection unit 30 can efficiently select music 120 that is suitable for the content of the audio-related information 50.

[0100] (5) The music selection unit 30 selects music 120 related to the matching word 6 and additional information as selected music 121 based on the matching word 6 and additional information and the related data 60. With this configuration, music 120 can be selected based on the audio related information 50 and additional information. Therefore, the music selection unit 30 can select music 120 more efficiently.

[0101] (6) Additional information includes the expected weather at the time of use of the sound source, the season at the time of use of the sound source, information about the day of use of the sound source, the location at the time of use of the sound source, visual information of people at the location at the time of use of the sound source, the purpose of use of the sound source, the sound environment of the location at which the sound source is used, the style of the store at which the sound source is used, the philosophy behind use of the sound source, the facilities of the person or organization using the sound source, personal information of people who may hear the sound source, and at least one additional keyword that can be set at will. With this configuration, not only audio-related information 50 but also music 120 can be selected based on this information.

[0102] (7) The related data 60 is Table 130. Table 130 is data in which related words 125 are associated with each of the multiple songs 120. With this configuration, songs 120 can be selected based on Table 130.

[0103] (8) The audio-related information 50 includes at least one of the announcement text information 50A, the announcement voice information, and the voice information. With this configuration, music 120 suitable for the audio-related information 50, which includes at least one of the announcement text information 50A, the announcement voice information, and the voice information, can be selected.

[0104] (9) The sound source formation program includes a word matching step of matching the related word 125 in relational data 60, which associates music 120 with related word 125, with the word in audio related information 50; a music selection step of forming a sound source by combining the selected music 121 with the audio data 122 relating to the audio related information 50. The sound source formation program causes a computer to perform these steps. In the music selection step, the computer selects music 120 related to the matching word 6 as the selected music 121, based on the matching word 6 and relational data 60 that were matched by the word matching step. With this configuration, the word matching step matches the related word 125 in relational data 60 with the word in audio related information 50. Then, the music selection step can select music 120 related to the matching word 6 obtained by the matching. In this way, the work of selecting music 120 suitable for the content of audio related information 50 can be reduced. This reduces the time required to form a sound source.

[0105] (10) The sound source formation method includes a word matching step of matching the related word 125 in relational data 60, which is associated with music 120 and related word 125, with the word in audio related information 50; a music selection step of forming a sound source by combining audio data 122 related to audio related information 50 and selected music 121. In the music selection step, music 120 related to the matching word 6, which was matched by the matching in the word matching step, is selected as selected music 121 based on the matching word 6 and relational data 60. With this configuration, the word matching step matches the related word 125 in relational data 60 with the word in audio related information 50. Then, the music selection step can select music 120 related to the matching word 6 obtained by the matching. In this way, the work of selecting music 120 suitable for the content of audio related information 50 can be reduced. This reduces the time required to form a sound source.

[0106] <Second Embodiment> The sound source formation system 1 of the second embodiment will now be described. In this embodiment, components common to the first embodiment are denoted by the same reference numerals as in the first embodiment, and the description of redundant components is omitted.

[0107] The relational data 60 of this embodiment differs from the relational data 60 of the first embodiment in the following respects. In the first embodiment, the relational data 60 is table 130. In contrast, in this embodiment, the relational data 60 is trained data. The trained model is formed by deep learning using a neural network.

[0108] The trained data is data obtained by training on a dataset of pre-associated music 120 and related words 125. For example, the trained data is composed of a trained model that has been trained using deep learning on a dataset of pre-associated music 120 and related words 125. In the trained data, multiple related words 125 and multiple music 120 are related via a neural network. The music selection unit 30 selects the music 120 that is most related to the concept containing multiple matching words 6 by inputting multiple matching words 6 into the trained data. This makes it easier for the music selection unit 30 to reduce the time required for selection. In addition, since the selection of keywords 21 is unnecessary, the chances of missing music 120 due to the selection of keywords 21 can be reduced.

[0109] <Variation> The embodiments described above are examples of possible forms of the sound source formation system 1, sound source formation program, and sound source formation method, and are not intended to limit their forms. The sound source formation system 1, sound source formation program, and sound source formation method may take forms different from those exemplified in the embodiments described above. One example of such a form is one in which a part of the configuration of the embodiment is replaced, modified, or omitted, or a form in which a new configuration is added to the embodiment. An example of a modified embodiment is shown below.

[0110] <First variation> As shown in Figure 10, the sound source may be configured as follows: The entire sound source is configured to start with a chime 123 and then fade in with the selected music 121A. The middle section of the sound source is configured so that the selected music 121A plays for a while, and then its volume gradually decreases. After the middle section of the sound source, the volume of the selected music 121A is reduced to a predetermined level, then the volume of the selected music 121A is maintained, and the first audio data 200 is played. After the first audio data 200 has finished playing, the second audio data 201 is played a short time later. After the second audio data 201 has finished playing, the selected music 121A is configured to gradually return to its original volume.

[0111] The first audio data 200 may be the same as the second audio data 201, or it may be different.

[0112] <Second variation> As shown in Figure 11, the sound source may be constructed by combining a first sound source 300 and a second sound source 301. The first sound source 300 may have the same data as the second sound source 301, or it may have different data.

[0113] <Third variation> • Referring to Figures 12 and 13, Table 130 of the fourth modified example will be described. Table 130 of this modified example includes a combination table. The combination table consists of the third table 133 and the fourth table 134.

[0114] As shown in Figure 12, Table 3 133 shows the relationship between facility name 111 and the information related to facility name 111. Facility name 111 is included in related word 125. In Table 3 133, each facility name 111 is linked to the information related to facility name 111.

[0115] As shown in Figure 13, Table 4, Table 134, shows the relationship between music 120 and information related to music 120. Music 120 is linked to information related to music 120. The music 120 shown in Table 4, Table 134 are Music 5, Music 6, Music 7, Music 8, Music 38.

[0116] According to Table 333, each facility name 111 is linked to information about that facility name 111. Specifically, facility name 111 is linked to the style of facility name 111, the genre of music to be played, and the sound environment of the site. On the other hand, in Table 434, each piece of music 120 is linked to information about that piece of music 120. Specifically, music 120 is linked to the genre, atmosphere, and tempo. As a result, the combination table allows the music selection unit 30 to indirectly select music 120 that suits each facility name 111 based on Table 333 and Table 434. For example, if facility name 111 is an event venue, the genre of music to be played is rock. The music selection unit 30 selects music 120 that suits the event venue from Table 434.

[0117] The music selection unit 30 then selects a piece of music 120 that is associated with the keyword 21 from among the music 120 selected by referring to the third table 133 and the fourth table 134. This selects music 120 that is suitable for the audio-related information 50.

[0118] <Fourth variation> Referring to Figure 14, the range setting unit for setting the range for matching words in the audio-related information 50 will be explained.

[0119] Referring to Figure 14, an example of the range for word matching is explained. The range setting unit sets the range for word matching in the audio-related information 50. In this example, the audio-related information 50 is the announcement text information 50A. The range setting unit sets the range for word matching to the part of the announcement text information 50A excluding "Welcome. We would like to inform our customers who are currently visiting our store." and "Thank you very much for visiting us today." The excluded part is the standard part. Based on the relationship data 60, the word matching unit 10 matches the related word 125 in the audio-related information 50 within the range set by the range setting unit, and then outputs the related word 125 that matches the matching of the word matching unit 10 as the matched word 6.

[0120] This range setting unit allows the user to set a range for matching words from the announcement text information 50B. This reduces the time required for the sound source formation system 1 to select the matching word 6.

[0121] Furthermore, if the audio-related information 50 is announcement audio information and audio information, the range setting unit may set the range for word matching as follows: The range setting unit sets the range for word matching based on the intonation in the announcement audio information and audio information. As an example, the range setting unit sets the range for matching the parts of the announcement audio information and audio information that include parts where the phonemes are uttered at a slower rate compared to other parts. This allows the word matching unit 10 to match words from important parts and parts that include important parts in the announcement audio information and audio information.

[0122] <Fifth variation> The word matching unit 10 may match the words in the audio-related information 50 with the related words 125 based on the words in the audio-related information 50. The words in the audio-related information 50 are words. Specifically, the types of words matched by the word matching unit 10 are nouns, words consisting of numerals and classifiers, and interjections. If the audio-related information 50 is announcement text information 50A, the word matching unit 10 matches nouns, words consisting of numerals and classifiers, and interjections from the announcement text information 50A with the related words 125 by morphological analysis. If the audio-related information 50 is audio data, the word matching unit 10 converts the audio-related information 50 into text information and then performs morphological analysis. [Explanation of symbols]

[0123] 1...Sound source formation system, 2...Network, 3...Server, 4...Input unit, 5...Output unit, 6...Matching word, 10...Word matching unit, 20...Keyword selection unit, 21...Keyword, 22...Priority data, 30...Music selection unit, 40...Synthesis unit, 50...Speech-related information, 50A...Announcement text information, 50B...Announcement text information, 60...Relationship data, 110...Registered word, 111...Facility name, 120...Music, 121...Selected music, 122...Speech data, 125...Related word, 130...Table, 131...First table, 132...Second table, 133...Third table, 134...Fourth table.

Claims

1. A sound source formation system for forming a sound source, A word matching unit compares the related words in the relational data with the words in the audio-related information by referring to audio-related information containing multiple words and relational data in which music and related words are associated. A keyword selection unit selects a keyword from matching words obtained by matching the related words of the related words and the words of the audio-related information by the word matching unit, A music selection unit that selects music related to the keyword as selected music based on the keyword and the related data, The system includes a synthesis unit that forms a sound source by combining audio data relating to the aforementioned audio-related information with the selected music, The aforementioned audio-related information is to be transmitted to the public or visitors and includes the message composed of the aforementioned multiple words, The aforementioned related words in the related data are words selected from a wide variety of communication items. The synthesis unit, based on a preset chime, the selected music, and the audio data, starts with the chime, and in the middle of the sound source, the volume of the selected music is lowered to a predetermined level while the audio data is played back, and after the playback of the audio data is finished, the volume of the selected music is returned to its original level, and in the latter part of the sound source, the chime is played back, thus forming the sound source. Sound source formation system.

2. A sound source formation system for forming a sound source, A word matching unit compares the related words in the relational data with the words in the audio-related information by referring to audio-related information containing multiple words and relational data in which music and related words are associated. A keyword selection unit selects a keyword from matching words obtained by matching the related words of the related words and the words of the audio-related information by the word matching unit, A music selection unit that selects music related to the keyword as selected music based on the keyword and the related data, The system includes a synthesis unit that forms a sound source by combining audio data relating to the aforementioned audio-related information with the selected music, The keyword selection unit selects the keyword from the matching words by referring to priority data. The aforementioned priority data includes multiple registered words, Each of the multiple registered words is assigned a priority level indicating its superiority as a keyword. Sound source formation system.

3. The aforementioned related data is, It is a table, The aforementioned table contains data in which the aforementioned related words are associated with each of several songs. The sound source formation system according to claim 1.

4. The aforementioned audio-related information includes at least one of announcement text information, announcement audio information, and audio information. The sound source formation system according to claim 1 or 2.

5. A sound source generation program for causing a computer to generate sound sources, A word matching step involves comparing the related words in the relational data with the words in the audio-related information by referring to audio-related information containing multiple words and relational data in which music and related words are associated. A keyword selection step involves selecting a keyword from the matching words obtained by matching the related words in the word matching step with the words in the audio related information, A music selection step that selects music related to the keyword based on the keyword and the related data, A synthesis step of forming a sound source by combining the audio data relating to the aforementioned audio-related information and the selected music, The computer is made to execute the above, The aforementioned audio-related information is to be transmitted to the public or visitors and includes the message composed of the aforementioned multiple words, The aforementioned related words in the related data are words selected from a wide variety of communication items. In the synthesis step, based on a pre-set chime, the selected music, and the audio data, the sound source is formed such that the sound source starts with the chime, and in the middle of the sound source, the volume of the selected music is reduced to a predetermined level while the audio data is maintained, and after the playback of the audio data is finished, the volume of the selected music is returned to its original level, and the end of the sound source plays the chime. Sound source generation program.

6. A sound source formation method for a sound source formation system, wherein The sound source formation system includes a word matching step of comparing the related words in the relational data with the words in the sound-related information by referring to speech-related information that includes multiple words and relational data that associates music with related words, The sound source formation system includes a keyword selection step, in which it selects a keyword from matching words that match the related words in the word matching step and the words in the sound-related information, The sound source formation system includes a music selection step of selecting music related to the keyword based on the keyword and the related data, The sound source formation system includes a synthesis step of forming a sound source by synthesizing audio data relating to the audio-related information and the selected music, The aforementioned audio-related information is to be transmitted to the public or visitors and includes the message composed of the aforementioned multiple words, The aforementioned related words in the related data are words selected from a wide variety of communication items. The sound source formation system, in the synthesis step, forms the sound source such that it starts with the chime, in the middle of the sound source the volume of the selected music is reduced to a predetermined level while the volume of the selected music is maintained, the audio data is played, after the playback of the audio data is finished the volume of the selected music is returned to its original level, and in the latter part of the sound source the chime is played. Sound source formation method.

Images