Information processing device, information processing method, program, and recording medium

Abstract

The present invention provides an information processing device which, even immediately after the information processing device is turned on, can determine a filter property for removing noise and can remove said noise.　The information processing device comprises: a filter property calculation unit 30 for calculating, on the basis of sound that is picked up, a filter property for removing noise from the sound; a filter unit 20 for removing noise from the picked up sound on the basis of the filter property calculated by the filter property calculation unit 30; a sound pickup environment detection unit 50 for detecting the sound pickup environment of the sound on the basis of a sensor unit 40; and a filter control unit 60 for acquiring the sound pickup environment at the time of startup and for setting a filter property for the filter unit 20 on the basis of the sound pickup environment at the time of start up and a filter table in which a sound pickup environment and the filter property calculated by the filter property calculation unit 30 are associated.

Description

[Technical Field]

[0001] The present invention relates to an information processing device, an information processing method, a program, and a recording medium. [Background technology]

[0002] In recent years, devices such as smartphones and smart speakers have become widespread, allowing users to activate voice assistants by speaking a wake word (also known as a wake-up word or hot word) and providing operational instructions, searching for information, and other functions based on the user's speech. Generally, when this type of device is operated in an environment containing noise, it may not be possible to correctly detect words spoken by the user. When this type of device is installed in a vehicle, the volume of noise and other disturbances varies greatly depending on the driving conditions (driving speed, engine speed, etc.) and vehicle conditions (air conditioner operation status, window opening / closing status, etc.). Therefore, technology has been disclosed that removes noise from the captured audio, making it easier to detect the words spoken by the user (see, for example, Patent Document 1). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2009-210647 Summary of the Invention [Problem to be solved by the invention]

[0004] In the above-mentioned prior art, a user's speech and a voice containing noise are picked up, and filter characteristics for removing the noise are determined based on the picked up voice. However, in the above-mentioned prior art, immediately after the device is turned on, no audio is picked up to determine the filter characteristics, so appropriate filter characteristics for removing noise cannot be determined, and the words spoken by the user cannot be correctly detected, as an example of the problem.

[0005] The present invention has been made in consideration of the problems cited as examples above, and has as its main object to provide an information processing device, an information processing method, a program, and a recording medium that determine optimal filter characteristics for removing noise and remove noise even immediately after the device is turned on. [Means for solving the problem]

[0006] In order to solve the above problem, the invention of claim 1 provides a sound collection device including: a filter characteristic calculation unit that calculates a filter characteristic for removing noise from a collected sound based on the collected sound; a filter unit that removes noise from the collected sound based on the filter characteristic calculated by the filter characteristic calculation unit; and a sound collection environment detection unit that detects a sound collection environment of the sound based on sensor information. a filter control unit that sets the filter characteristics calculated by the filter characteristics calculation unit to the filter unit during a period other than the start-up period; Equipped with The filter control unit At startup only In the above, the sound collection environment is acquired, and the sound collection environment at the time of startup is acquired. in advance The filter characteristics are set in the filter unit based on a filter table that stores the filter characteristics calculated by the filter characteristics calculation unit in association with the sound collection environment. Characterized by It is an information processing device.

[0007] Also, claims 6The invention described in is an information processing method in an information processing device including a filter characteristic calculation unit, a filter unit, a sound collection environment detection unit, and a filter control unit, the information processing method including a first step in which the filter characteristic calculation unit calculates filter characteristics for removing noise from a sound based on the sound collected, a second step in which the filter unit removes noise from the sound collected based on the filter characteristics calculated by the filter characteristic calculation unit, a third step in which the sound collection environment detection unit detects the sound collection environment based on sensor information, and the filter control unit: and a fourth step of setting the filter characteristics calculated by the filter characteristics calculation unit in the filter unit during a period other than the start-up period, wherein in the fourth step, the filter control unit: At startup only In the above, the sound collection environment is acquired, and the sound collection environment at the time of startup is acquired. in advance The filter characteristics are set in the filter unit based on a filter table that stores the filter characteristics calculated by the filter characteristics calculation unit in association with the sound collection environment. Characterized by It is an information processing method.

[0008] Also, claims 7 The invention described in is a program for causing a computer to execute an information processing method in an information processing device including a filter characteristic calculation unit, a filter unit, a sound collection environment detection unit, and a filter control unit, the program including a first step in which the filter characteristic calculation unit calculates filter characteristics for removing noise from a sound based on the sound collected, a second step in which the filter unit removes noise from the sound collected based on the filter characteristics calculated by the filter characteristic calculation unit, a third step in which the sound collection environment detection unit detects the sound collection environment based on sensor information, and the filter control unit and a fourth step of setting the filter characteristics calculated by the filter characteristics calculation unit in the filter unit during a period other than the start-up period, wherein in the fourth step, the filter control unit: At startup only In the above, the sound collection environment is acquired, and the sound collection environment at the time of startup is acquired. in advance The program causes a computer to execute an information processing method for setting the filter characteristics in the filter unit based on a filter table that stores the filter characteristics calculated by the filter characteristics calculation unit in association with the sound collection environment.

[0009] Also, claims 8The invention described in is a non-transitory computer-readable recording medium having recorded thereon a program for causing a computer to execute an information processing method in an information processing device including a filter characteristic calculation unit, a filter unit, a sound collection environment detection unit, and a filter control unit, the information processing method including a first step in which the filter characteristic calculation unit calculates filter characteristics for removing noise from a sound based on the sound collected, a second step in which the filter unit removes noise from the sound collected based on the filter characteristics calculated by the filter characteristic calculation unit, a third step in which the sound collection environment detection unit detects the sound collection environment based on sensor information, and the filter control unit: and a fourth step of setting the filter characteristics calculated by the filter characteristics calculation unit in the filter unit during a period other than the start-up period, wherein in the fourth step, the filter control unit: At startup only In the above, the sound collection environment is acquired, and the sound collection environment at the time of startup is acquired. in advance The recording medium stores a program for causing a computer to execute an information processing method for setting the filter characteristics in the filter unit based on a filter table that stores the filter characteristics calculated by the filter characteristics calculation unit and the sound collection environment in association with each other. [Brief explanation of the drawings]

[0010] [Figure 1] 1 is a diagram illustrating a configuration of an information processing device according to an embodiment of the present invention. [Figure 2] 3 is a diagram illustrating a filter table generated and referenced by a filter control unit of the information processing device according to the embodiment of the present invention; FIG. [Figure 3] FIG. 4 is a diagram showing a processing flow of a filter control unit of the information processing device according to the embodiment of the present invention. [Figure 4] 10 is a diagram illustrating a process performed when a filter control unit of the information processing device according to the embodiment of the present invention refers to a filter table; [Figure 5] 10A and 10B are diagrams illustrating processing when a filter control unit of the information processing device according to the embodiment of the present invention adds a sound collection environment and a filter characteristic to a filter table. [Figure 6] FIG. 10 is a diagram illustrating sensor information acquired by a sensor unit of an information processing device according to another embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0011] The information processing device according to this embodiment includes a filter characteristic calculation unit that calculates filter characteristics for removing noise from audio based on collected audio, a filter unit that removes noise from the collected audio based on the filter characteristics calculated by the filter characteristic calculation unit, a sound collection environment detection unit that detects the audio collection environment based on sensor information, and a filter control unit that acquires the sound collection environment at startup and sets the filter characteristics in the filter unit based on the sound collection environment at startup and a filter table that links the filter characteristics calculated by the filter characteristic calculation unit with the sound collection environment. The filter characteristic calculation unit calculates filter characteristics for removing noise from the audio based on the picked-up audio. The filter unit removes noise from the collected voice based on the filter characteristics calculated by the filter characteristics calculation unit. The sound collection environment detection unit detects the sound collection environment in which sound is collected based on camera images, sensor information from vehicle sensors, and the like. The filter control unit acquires the sound collection environment when the information processing device is started up, and determines the filter characteristics to be set in the filter unit based on the sound collection environment at the time of startup and a filter table that links the filter characteristics calculated by the filter characteristic calculation unit with the sound collection environment, and sets the filter characteristics in the filter unit. If the sound collection environment at the time of startup of the information processing device is in the filter table, the filter control unit sets the filter characteristics linked to the sound collection environment at the time of startup in the filter unit, and if the sound collection environment is not in the filter table, sets the filter characteristics last set in the filter unit in the filter unit. When the information processing device is started up, no sound is collected for determining the filter characteristics, and therefore it is not possible to determine appropriate filter characteristics for removing noise. Therefore, the filter control unit determines the filter characteristics to be set in the filter unit based on the sound collection environment at the time of startup and the filter table. This allows the filter characteristics to be set based on the sound collection environment at the time of startup, even when no sound is collected to determine the filter characteristics, so that noise can be appropriately removed from the sound. Further, the filter control unit sets the filter characteristics calculated by the filter characteristics calculation unit in the filter unit during a period other than when the information processing device is started up. That is, except when the information processing device is started up, the filter characteristics calculated by the filter characteristics calculation unit based on the sound picked up by the sound pickup unit are set in the filter unit. As a result, during the period other than the time when the information processing device is started up, the filter characteristics are calculated based on the sound in the space where the user speaks, so that the optimum filter characteristics can be set.

[0012] Furthermore, even if the sound collection environment at the time of startup is not included in the filter table, the filter characteristics last set in the filter section are set, so that better filter characteristics can be set for the space in which the user speaks. The filter characteristics last set in the filter unit as mentioned above refer to the filter characteristics set in the filter unit when the information processing device is stopped from operating due to power-off or the like, and when the information processing device is next started and the sound collection environment at the time of startup is not in the filter table, the filter characteristics are set in the filter unit. In other words, even when the information processing device is started up, proven filter characteristics calculated based on the sound of the space in which the user is speaking are set, so that more appropriate filter characteristics can be set for that space. Furthermore, the filter control unit acquires the sound collection environment when the filter characteristics are calculated by the filter characteristic calculation unit, and if the sound collection environment is not in the filter table, the sound collection environment is linked to the filter characteristics calculated by the filter characteristic calculation unit and added to the filter table. This allows the optimal filter characteristics for each sound collection environment to be stored in the filter table simply by operating the information processing device, so that the filter control unit can set the optimal filter characteristics in the filter unit at startup by referring to the filter table.

[0013] <Example> An information processing device 1 according to this embodiment will be described with reference to FIGS.

[0014] <Configuration of information processing device 1> The configuration of an information processing device 1 according to this embodiment will be described with reference to FIG. The information processing device 1 includes at least a sound collection unit 10, a filter unit 20, a filter characteristic calculation unit 30, a sensor unit 40, a sound collection environment detection unit 50, and a filter control unit 60.

[0015] The sound collection unit 10 is configured by, for example, a microphone, collects sound from within the vehicle cabin, and transmits the collected sound to the filter unit 20 and the filter characteristic calculation unit 30. The sound collected by the sound collection unit 10 includes the user's speech, as well as noise and other disturbances occurring around the microphone. Specifically, the sounds picked up by the microphone installed inside the vehicle cabin include engine noise, wind noise, road noise, air conditioner operation sounds, music output from the speakers, etc. while driving. The microphone may be a microphone installed in the vehicle, for example, a microphone for hands-free calling, as long as it can pick up the above-mentioned sounds inside the vehicle.

[0016] The filter unit 20 removes noise from the sound collected by the sound collection unit 10 based on the filter characteristics received from the filter control unit 60, which will be described later. The noise-removed voice is input to a voice recognition engine (not shown) to detect the words spoken by the user.

[0017] The filter characteristic calculation unit 30 calculates, based on the sound collected by the sound collection unit 10, filter characteristics for removing noise from the sound. Specifically, the filter characteristic calculation unit 30 divides the sound collected by the sound collection unit 10 into pieces of sound data, for example, for 20 seconds each, and calculates filter characteristics for removing noise for each piece of divided sound data. The filter characteristics calculated by the filter characteristics calculation unit 30 are transmitted to the filter control unit 60, which will be described later.

[0018] The sensor unit 40 is configured with at least a camera that takes images inside the vehicle and a sensor that detects the state of the vehicle, and transmits the acquired sensor information to the sound collection environment detection unit 50, which will be described later. Here, the image transmitted from the sensor unit 40 as sensor information only needs to be an image of the inside of the vehicle, so it is possible to transmit to the sound collection environment detection unit 50 an image captured by a drive recorder installed in the vehicle. Examples of sensor information for detecting the state of the vehicle include vehicle speed pulses, acceleration sensors, GPS signals, and information from various sensors connected to an ECU (Electronic Control Unit) of the vehicle.

[0019] The sound collection environment detection unit 50 detects the sound collection environment based on the sensor information from the sensor unit 40. Specifically, the sound collection environment detection unit 50 analyzes images captured inside the vehicle, and detects, for example, the passenger's riding position, the passenger's gender, and whether the vehicle windows are open or closed as the sound collection environment. Furthermore, the sound collection environment detection unit 50 detects the vehicle's running speed, engine rotation speed, air conditioner operation status, etc. as the sound collection environment based on sensor information indicating the vehicle's state. The sound collection environment detection unit 50 transmits the detected sound collection environment to the filter control unit 60. Furthermore, the sound collection environment detection unit 50 detects the sound collection environment when the filter characteristic calculation unit 30 calculates the filter characteristic, and transmits the detected sound collection environment to the filter control unit 60. For example, the sound collection environment detection unit 50 calculates the average value of the sound collection environment (average value of engine rotation speed, driving speed, etc.) during the period when the filter characteristics are calculated in the filter characteristic calculation unit 30, and transmits the sound collection environment to the filter control unit 60.

[0020] The filter control unit 60 acquires the sound collection environment at the time of startup, and sets the filter characteristics in the filter unit 20 based on the sound collection environment at the time of startup and a filter table that links the filter characteristics calculated by the filter characteristic calculation unit 30 with the sound collection environment when the filter characteristics were calculated. That is, when the information processing device 1 is started up, the sound for determining the filter characteristics has not yet been collected, and therefore the filter characteristics calculation unit 30 cannot calculate the filter characteristics. Therefore, the filter control unit 60 determines the filter characteristics to be set in the filter unit 20 based on the sound collection environment acquired at the time of startup and a filter table such as that shown in FIG.

[0021] The filter table stores the sound collection environment when the filter characteristics are calculated by the filter characteristics calculation unit 30 and the calculated filter characteristics in association with each other. Specifically, the sound collection environment (riding position, gender, window open / close status, driving speed, engine RPM, air conditioner operation status) detected from the camera image of the sensor unit 40 and vehicle sensor information is linked to the filter characteristics calculated by the filter characteristic calculation unit 30 and stored in the filter table. More specifically, as shown in FIG. 2, sound collection environments K1 to K5 and filter characteristics F1 to F5 are associated with each other and stored in the filter table.

[0022] If the filter table contains a sound collection environment that is the same as the sound collection environment acquired at startup, the filter control unit 60 acquires the filter characteristics associated with that sound collection environment from the filter table and sets the filter characteristics in the filter unit 20. On the other hand, if the filter table does not contain the same sound collection environment as the sound collection environment acquired at the time of startup, the filter characteristics last set in the filter section 20 are set in the filter section 20. The filter characteristics last set in the filter unit 20 mentioned above are the filter characteristics that were set in the filter unit 20 when the information processing device 1 was stopped from operating due to power-off or the like, and when the information processing device 1 is started next time and the sound collection environment acquired at the time of startup is not in the filter table, the filter characteristics are set in the filter unit 20. The filter control unit 60 stores the filter characteristic values that were last set in the filter unit 20 in a memory (not shown).

[0023] Furthermore, the filter control unit 60 sets the filter characteristics calculated by the filter characteristics calculation unit 30 in the filter unit 20 during periods other than the start-up period.

[0024] The filter control unit 60 acquires the sound collection environment when the filter characteristics are calculated by the filter characteristics calculation unit 30 from the sound collection environment detection unit 50, and if the sound collection environment is not in the filter table, associates the sound collection environment with the filter characteristics and adds it to the filter table. The processing performed by the filter control unit 60 will be described in detail below.

[0025] <Processing of the filter control unit 60> The processing of the filter control unit 60 will be described in detail with reference to FIGS.

[0026] As shown in FIG. 3, it is determined whether or not the ACC power supply (accessory power supply) of the vehicle is in an ON state (step S100). If it is determined that the ACC power supply of the vehicle is not in an ON state ("NO" in step S100), the process returns to step S100 and transitions to a standby state. On the other hand, if it is determined that the ACC power supply of the vehicle is in the ON state ("YES" in step S100), the process proceeds to step S110.

[0027] If it is determined that the ACC power supply of the vehicle is in an on state ("YES" in step S100), the sound collection environment is acquired from the sound collection environment detection unit 50 (step S110). That is, the filter control unit 60 acquires the current sound collection environment from the sound collection environment detection unit 50 immediately after the ACC power supply is turned on (the information processing device 1 is turned on).

[0028] The filter control unit 60 determines whether the sound collection environment acquired in step S110 is in the filter table (step S120). If it is determined that the acquired sound collection environment is in the filter table ("YES" in step S120), the process proceeds to step S130. On the other hand, if it is determined that the acquired sound collection environment is not in the filter table ("NO" in step S120), the process proceeds to step S140.

[0029] Here, a method for determining whether or not the same sound collection environment as the sound collection environment acquired from the sound collection environment detection unit 50 exists in the filter table will be described. The information indicating the sound collection environment acquired from the sound collection environment detection unit 50 includes sound collection environment values whose values change significantly during driving, such as the driving speed and engine rotation speed. Therefore, when determining whether or not the filter table contains a sound collection environment that is the same as the sound collection environment received from the sound collection environment detection unit 50, a similarity is calculated to determine whether or not they are the same.

[0030] Using FIG. 4, the similarity will be described by exemplifying a case where the value of the sound collection environment KA obtained immediately after startup is compared with the sound collection environments K1 to K3 in the filter table. The filter control unit 60 determines whether the value of the traveling speed in the sound collection environment at the time of startup, which is acquired from the sound collection environment detection unit 50, is K A1 If so, the value of the travel speed (K 11 , K. 21 , K. 31 ) and calculate the absolute value of the difference, which is the similarity. If the calculated similarity is smaller than a predetermined value (for example, if the similarity is less than 10 Km / h), the filter control unit 60 determines that the two travel speed values are the same. Similarly, when the similarity calculated for engine speed is smaller than a predetermined value (for example, when the similarity is less than 200 rpm), the two engine speeds are determined to be the same numerical value.

[0031] In the above-mentioned similarity determination, if it is determined that both the driving speed and the engine rotation speed are the same, and further if it is determined that the other sound collection environments (riding position, gender, window opening / closing status, air conditioning operation status) are the same, the filter control unit 60 determines that the sound collection environments are the same (the sound collection environment KA and the sound collection environment K3 obtained from the sound collection environment detection unit 50 are determined to be the same sound collection environment). When the above-described similarity determination is performed and it is determined that the sound collection environments are the same, for example, the sound collection environment with the smallest similarity value is determined to be the same sound collection environment.

[0032] If it is determined that the sound collection environment acquired at startup (the sound collection environment acquired in step S110) is in the filter table ("YES" in step S120), the filter characteristics associated with that sound collection environment are acquired from the filter table, and the filter characteristics are set in the filter unit 20 (step S130).

[0033] On the other hand, if it is determined that the sound collection environment acquired at startup (the sound collection environment acquired in step S110) is not in the filter table ("NO" in step S120), the filter characteristics last set in the filter unit 20, which are stored in a memory not shown, are set in the filter unit 20 (step S140).

[0034] The filter control unit 60 acquires the filter characteristics calculated by the filter characteristics calculation unit 30, and sets the acquired filter characteristics in the filter unit 20 (step S150). In other words, once a predetermined time has passed since startup, the filter characteristic calculation unit 30 can calculate the filter characteristics, and the filter control unit 60 sets the filter characteristics calculated by the filter characteristic calculation unit 30 to the filter unit 20 during periods other than startup.

[0035] The filter control unit 60 acquires the sound collection environment at the time when the filter characteristics set in the filter unit 20 were calculated in step S150 from the sound collection environment detection unit 50 (step S160).

[0036] The filter control unit 60 stores the filter characteristics set in the filter unit 20 in a memory (not shown) (step S170). That is, in step S170, the process of storing the filter characteristic values last set in the filter unit 20 in memory is executed.

[0037] The filter control unit 60 determines whether or not the same sound collection environment as the sound collection environment acquired in step S160 exists in the filter table (step S180). If it is determined that the same sound collection environment exists in the filter table ("YES" in step S180), the process proceeds to step S200. On the other hand, if it is determined that the same sound collection environment is not in the filter table ("NO" in step S180), the filter characteristics set in the filter unit 20 in step S150 are linked to the sound collection environment acquired in step S160 and added to the filter table (step S190).

[0038] That is, the filter control unit 60 acquires the sound collection environment when the filter characteristics were calculated by the filter characteristics calculation unit 30 from the sound collection environment detection unit 50, associates the sound collection environment with the filter characteristics, and adds them to the filter table. Specifically, the sound collection environment information obtained from the sound collection environment detection unit 50, such as the passenger's riding position, the gender of each passenger, the open / close state of the vehicle windows, the traveling speed, the engine speed, the air conditioner operation status, etc., is compared with the sound collection environment in the filter table to determine whether the same sound collection environment is in the filter table, and if the same sound collection environment is not in the filter table, the sound collection environment and the filter characteristics are linked and added to the filter table. More specifically, for example, as shown in FIG. 5, it is determined whether or not there is a sound collection environment K1 to K5 registered in the filter table that is the same as the sound collection environment acquired from the sound collection environment detection unit 50. If there is no such sound collection environment, the sound collection environment K6 and the filter characteristic F6 associated with the sound collection environment are added to the filter table as a new sound collection environment K6. The method of determining whether the same sound collection environment is in the filter table is the same as the determination method in step S120 described above.

[0039] The filter control unit 60 determines whether or not the ACC power supply (accessory power supply) of the vehicle is in an ON state (step S200). If it is determined that the ACC power supply of the vehicle is in the ON state ("YES" in step S200), the process proceeds to step S150 and continues. On the other hand, if it is determined that the ACC power supply of the vehicle is not in an ON state ("NO" in step S200), the process is ended.

[0040] The information processing device 1 according to this embodiment includes a filter characteristic calculation unit 30 that calculates filter characteristics for removing noise from a collected sound based on the sound, a filter unit 20 that removes noise from the collected sound based on the filter characteristics calculated by the filter characteristic calculation unit 30, a sound collection environment detection unit 50 that detects the sound collection environment based on a sensor unit 40, and a filter control unit 60 that acquires the sound collection environment at startup and sets the filter characteristics in the filter unit 20 based on a filter table that links the sound collection environment at startup with the filter characteristics calculated by the filter characteristic calculation unit 30 and the sound collection environment. The filter unit 20 removes noise from the collected voice based on the filter characteristics calculated by the filter characteristics calculation unit 30 . The filter characteristic calculation unit 30 calculates, based on the sound collected by the sound collection unit 10, filter characteristics for removing noise from the sound. The sound collection environment detection unit 50 detects the sound collection environment in which sound is collected based on information from the sensor unit 40 such as a camera image and a vehicle sensor. The filter control unit 60 acquires the sound collection environment when the information processing device 1 is started up, and determines the filter characteristics to be set in the filter unit 20 based on the sound collection environment at the time of startup and the filter table, and sets the filter characteristics in the filter unit 20. If the sound collection environment at the time of startup of the information processing device 1 is in the filter table, the filter control unit 60 sets the filter characteristics associated with the sound collection environment at the time of startup to the filter unit 20, and if the sound collection environment is not in the filter table, the filter control unit 60 sets the filter characteristics last set to the filter unit 20 to the filter unit 20. Immediately after startup, no sound is collected to determine the filter characteristics, so appropriate filter characteristics cannot be determined. Therefore, the filter control unit 60 determines the filter characteristics to be set in the filter unit 20 based on the sound collection environment at the time of startup and the filter table. This allows the filter characteristics to be set based on the sound collection environment at the time of startup, even when no sound is collected to determine the filter characteristics, so that noise can be appropriately removed from the sound. Furthermore, even if the sound collection environment at the time of startup is not present in the filter table, the filter characteristics last set in the filter unit 20 are set. That is, since proven filter characteristics calculated based on the sound of the space in which the user speaks are set, it is possible to set filter characteristics that are more appropriate for that space.

[0041] Furthermore, the filter control unit 60 acquires the sound collection environment when the filter characteristics are calculated by the filter characteristics calculation unit 30, and if the sound collection environment is not in the filter table, the filter control unit 60 associates the sound collection environment with the filter characteristics calculated by the filter characteristics calculation unit 30 and adds them to the filter table. The filter characteristics calculated by the filter characteristic calculation unit 30 vary greatly in value depending on the sound collection environment. For example, the volume and frequency of engine sounds contained in the collected audio change depending on the engine speed. Furthermore, the volume and frequency of road noise contained in the picked-up audio changes depending on the driving speed. Furthermore, the volume and frequency of the speech contained in the collected audio changes depending on the gender of the speaker. Therefore, by linking the sound pickup environment at the time the filter characteristics were calculated with the calculated filter characteristics and storing them in a filter table, it is possible to store the filter characteristics that are optimal for the vehicle and occupants. In other words, simply by operating the information processing device 1, the filter characteristics for each sound collection environment that are optimal for that space can be stored in the filter table, and the filter control unit 60 can set the optimal filter characteristics by referring to the filter table.

[0042] Furthermore, the filter control unit 60 sets the filter characteristics calculated by the filter characteristics calculation unit 30 in the filter unit 20 during periods other than when the information processing device 1 is started up. That is, except when the information processing device 1 is started up, the filter characteristic calculation unit 30 sets the filter characteristic calculated based on the sound collected by the sound collection unit 10 in the filter unit 20. As a result, except when the information processing device 1 is started up, the filter characteristics are calculated based on the sound in the space where the user speaks, so that the optimum filter characteristics can be set.

[0043] The sensor information of the sensor unit 40 includes at least an image captured inside the vehicle and information indicating the running state of the vehicle. That is, a filter table is created in association with information such as the passenger position, gender, window opening / closing status, driving speed, and engine RPM, which are factors that affect the filter characteristics for removing noise contained in the collected voice, and the filter values to be set in the filter unit 20 are determined based on the filter table and the sound collection environment at the time of startup. This allows the optimum filter characteristics for the space in which the user speaks to be determined based on the sound collection environment at the time of startup, making it possible to remove noise from the collected voice even at startup.

[0044] The sound collected by the sound collection unit 10 contains noise and the user's speech, and the sound from which the noise has been removed by the filter unit 20 is sent to a voice recognition engine. That is, under the control of the filter control unit 60, optimal filter characteristics are set in the filter unit 20 even immediately after the information processing device 1 is started up, so that noise can be removed from the sound collected by the sound collection unit 10. This makes it possible to improve the recognition rate of the spoken voice in the voice recognition engine.

[0045] <Other Examples> The sound collection environment detection unit 50 described above detects the passenger positions, the gender of each passenger, the open / closed state of the vehicle windows, the traveling speed, the engine speed, the air conditioner operation status, etc. as the sound collection environment, but may also detect further information such as that shown in FIG. 6. Specifically, the current weather conditions and the driving conditions of surrounding vehicles may be detected as the sound collection environment from images taken inside and outside the vehicle by a camera. Since road noise may increase during rainy weather compared to sunny days, more appropriate filter characteristics can be set by adding the current weather as a condition of the sound collection environment. In addition, if trucks, motorcycles, etc. are traveling around the vehicle, there is a possibility that the noise from the vehicles traveling around the vehicle may increase, so by detecting the state of the vehicles traveling around the vehicle as the sound collection environment, more appropriate filter characteristics can be set. Alternatively, the current location of the vehicle may be detected as the sound collection environment from GPS (Global Positioning System) information of the vehicle. For example, there is noise that occurs uniquely in driving locations such as highways, residential areas, and urban areas, so by knowing the driving location, more appropriate filter characteristics can be set.

[0046] Furthermore, the above-mentioned sound collection environment detection unit 50 detects the speaker's position, etc. based on images taken inside the vehicle, but it may also be possible to detect the passenger's riding position as the sound collection environment from the sound collected using a microphone array. This makes it possible to determine the passenger's position even if a camera cannot be installed to capture images inside the vehicle.

[0047] In addition, in the above-described information processing device 1, the filter table is generated in the filter control unit 60, but the sound collection environment received from the sound collection environment detection unit 50 may be transmitted to a server via an Internet line, and the filter table may be generated in the server. This eliminates the load of the filter table creation process on the filter control unit 60, thereby reducing power consumption. Furthermore, the memory capacity for storing and creating the filter table can be reduced or eliminated, so the cost of the information processing device 1 can be reduced. Furthermore, by storing the filter table on a server, the filter table can be shared with other users. Specifically, for example, the server aggregates and analyzes data from filter tables for the same vehicle type, and generates a filter table that can be shared for each vehicle type. This allows the filter characteristics to be set by referring to the shared filter table, so that optimal filter characteristics can be set even when the information processing device 1A is used for the first time or immediately after startup.

[0048] In the above-described information processing device 1, if the sound collection environment acquired at startup is not in the filter table, the filter characteristics last set in the filter unit 20 are set in the filter unit 20, but it is also possible to set the filter characteristics associated with the sound collection environment that is closest to the sound collection environment acquired at startup in the filter unit 20. Specifically, for example, even if the above-mentioned similarity is equal to or greater than a predetermined value, the sound collection environment showing the smallest similarity value is determined to be the same sound collection environment, and the filter characteristics associated with that sound collection environment are set in the filter unit 20. This allows the filter section 20 to be set with filter characteristics associated with the sound collection environment that is closest to the sound collection environment acquired at startup, making it possible to set optimal filter characteristics for the space in which the user speaks.

[0049] Although an embodiment of the present invention has been described above in detail with reference to the drawings, the specific configuration is not limited to this embodiment, and includes designs within the scope of the gist of the present invention. [Explanation of symbols]

[0050] 1. Information processing device 10; Sound pickup section 20: Filter section 30: Filter characteristic calculation section 40: Sensor section 50; Sound collection environment acquisition section 60: Filter control section

Claims

1. a filter characteristic calculation unit that calculates a filter characteristic for removing noise from a collected voice based on the collected voice; a filter unit that removes noise from the collected voice based on the filter characteristics calculated by the filter characteristics calculation unit; a sound collection environment detection unit that detects a sound collection environment of the sound based on sensor information; a filter control unit that sets the filter characteristics calculated by the filter characteristics calculation unit to the filter unit during a period other than the start-up period; Equipped with the filter control unit acquires the sound collection environment only at startup, and sets the filter characteristics in the filter unit based on the sound collection environment at startup and a filter table that associates the filter characteristics calculated in advance by the filter characteristic calculation unit with the sound collection environment. An information processing device comprising:

2. 2. The information processing device according to claim 1, wherein, if the sound collection environment at the time of startup is in the filter table, the filter control unit sets the filter characteristics associated with the sound collection environment at the time of startup to the filter unit, and if the sound collection environment is not in the filter table, the filter control unit sets the filter characteristics last set to the filter unit to the filter unit.

3. 3. The information processing apparatus according to claim 1, wherein the filter control unit acquires the sound collection environment when the filter characteristics are calculated by the filter characteristics calculation unit, and when the sound collection environment is not in the filter table, the sound collection environment and the filter characteristics calculated by the filter characteristics calculation unit are linked to each other and added to the filter table.

4. 2. The information processing apparatus according to claim 1, wherein the sensor information includes at least an image captured inside the vehicle and information indicating a running state of the vehicle.

5. 2. The information processing apparatus according to claim 1, wherein the collected voice includes the noise and a voice uttered by the user, and the voice from which the noise has been removed by the filter unit is transmitted to a voice recognition engine.

6. An information processing method in an information processing device including a filter characteristic calculation unit, a filter unit, a sound collection environment detection unit, and a filter control unit, a first step in which the filter characteristic calculation unit calculates a filter characteristic for removing noise from the audio based on the audio; a second step in which the filter unit removes noise from the collected voice based on the filter characteristics calculated by the filter characteristics calculation unit; a third step in which the sound collection environment detection unit detects the sound collection environment based on sensor information; a fourth step in which the filter control unit sets the filter characteristics calculated by the filter characteristics calculation unit in the filter unit during a period other than a start-up period; Equipped with the fourth step is an information processing method characterized in that the filter control unit acquires the sound collection environment only at startup, and sets the filter characteristics in the filter unit based on the sound collection environment at startup and a filter table that associates the filter characteristics calculated in advance by the filter characteristic calculation unit with the sound collection environment.

7. A program for causing a computer to execute an information processing method in an information processing device including a filter characteristic calculation unit, a filter unit, a sound collection environment detection unit, and a filter control unit, a first step in which the filter characteristic calculation unit calculates a filter characteristic for removing noise from the audio based on the audio; a second step in which the filter unit removes noise from the collected voice based on the filter characteristics calculated by the filter characteristics calculation unit; a third step in which the sound collection environment detection unit detects the sound collection environment based on sensor information; a fourth step in which the filter control unit sets the filter characteristics calculated by the filter characteristics calculation unit in the filter unit during a period other than a start-up period; Equipped with In the fourth step, the filter control unit acquires the sound collection environment only at startup, and sets the filter characteristics in the filter unit based on the sound collection environment at startup and a filter table that stores the filter characteristics calculated in advance by the filter characteristic calculation unit in association with the sound collection environment.

8. A non-transitory computer-readable recording medium storing a program for causing a computer to execute an information processing method in an information processing device including a filter characteristic calculation unit, a filter unit, a sound collection environment detection unit, and a filter control unit, a first step in which the filter characteristic calculation unit calculates a filter characteristic for removing noise from the audio based on the audio; a second step in which the filter unit removes noise from the collected voice based on the filter characteristics calculated by the filter characteristics calculation unit; a third step in which the sound collection environment detection unit detects the sound collection environment based on sensor information; a fourth step in which the filter control unit sets the filter characteristics calculated by the filter characteristics calculation unit in the filter unit during a period other than a start-up period; Equipped with a recording medium having a program recorded thereon for causing a computer to execute an information processing method in which, in the fourth step, the filter control unit acquires the sound collection environment only at startup, and sets the filter characteristics in the filter unit based on the sound collection environment at startup and a filter table that associates and stores the filter characteristics calculated in advance by the filter characteristic calculation unit with the sound collection environment.

Images