Automatic broadcast sound adjustment device, automatic broadcast sound adjustment method, and program

Abstract

To be possible to generate an automatic broadcast audio signal that is easy to listen to.SOLUTION: There is provided an automatic broadcast sound conditioning device including a control unit that adjusts a signal waveform of an automatic broadcast sound signal according to a speech recognition rate based on a comparison between a first text data used for the automatic broadcast and a second text data generated from the sound signal obtained by sound collection of broadcast speech based on the automatic broadcast sound signal generated from the first text data.SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] The present invention relates to an automatic broadcast sound adjustment device, an automatic broadcast sound adjustment method, and a program. [Background technology]

[0002] Conventionally, there is an automatic announcement device for elevators that adjusts the volume (see, for example, Patent Document 1). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] International Publication No. WO2006 / 006204 Summary of the Invention [Problem to be solved by the invention]

[0004] An object of the present invention is to provide an automatic broadcast sound adjustment device, an automatic broadcast sound adjustment method, and a program capable of generating an automatic broadcast audio signal that is easy to listen to. [Means for solving the problem]

[0005] One aspect of the present invention is an automatic broadcast sound adjustment device that includes a control unit that calculates a voice recognition rate by comparing first text data used for automatic broadcasting with second text data generated from an audio signal obtained by collecting broadcast audio based on an automatic broadcast audio signal generated from the first text data, and adjusts the signal waveform of the automatic broadcast audio signal in accordance with the voice recognition rate.

[0006] According to the automatic broadcast sound adjustment device, the signal waveform of the automatic broadcast sound signal is adjusted according to the voice recognition rate, making it possible to generate an automatic broadcast sound that is easy to listen to.

[0007] In the automatic broadcast sound adjustment device, it is preferable that the adjustment section is configured to perform at least one of volume adjustment, frequency adjustment, compressor adjustment, and limiter adjustment on the automatic broadcast audio signal.

[0008] In addition, in the automatic broadcast sound adjustment device, it is preferable that the adjustment section adjusts the automatic broadcast audio signal based on a comparison result between a waveform of the audio signal and a waveform of the automatic broadcast audio signal.

[0009] In the automatic broadcast sound control device, the first text data is preferably text data for in-car broadcasting in a railway vehicle, but may be text data for automatic broadcasting in a moving body other than a railway vehicle. The first text data is preferably text data in a language other than Japanese, but may be Japanese text data.

[0010] Further, other aspects of the present invention may include an automatic broadcast sound adjustment method, a program, and a recording medium having the program recorded thereon, each of which has features similar to those of the automatic broadcast sound adjustment device. Effect of the Invention

[0011] According to the present invention, it is possible to generate an automatic broadcast audio signal that is easy to listen to. [Brief description of the drawings]

[0012] [Figure 1] FIG. 1 shows an example of an automatic public address system for a railway vehicle. [Diagram 2] FIG. 2 shows an example of the configuration of a terminal. [Diagram 3] FIG. 3 is a flowchart showing an example of processing in the terminal. [Figure 4] FIG. 4 is a flowchart showing an example of processing in the terminal. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0013] Hereinafter, an automatic broadcast sound adjustment device, an automatic broadcast sound adjustment method, and a program according to an embodiment will be described with reference to the drawings. The configurations of the embodiments are merely examples, and the present invention is not limited to the configurations of the embodiments. This embodiment shows an example in which the automatic broadcast sound adjustment device is applied to an automatic broadcast system of a railway vehicle. However, the automatic method sound adjustment device may also be applied to moving bodies other than railway vehicles (vehicles such as buses, ships, aircraft, etc.).

[0014] FIG. 1 shows an example of an automatic public address system 100 installed in a railway vehicle. In FIG. 1, the automatic public address system 100 automatically broadcasts audio according to the operating state of the railway vehicle. The public address system is roughly composed of a CPU (Central Processing Unit) 102, a storage device 103, and A sound signal generating unit 104, a switch (SW) 15, a control amplifier (C.AMP) 106, and an output A power amplifier (P.AMP) 107, a speaker 108, and a microphone 109 and a terminal 10. A speaker 108 and a microphone 109 are disposed in the passenger compartment.

[0015] The storage device 103 includes a main storage device such as a random access memory (RAM) or a read only memory (ROM), a hard disk, a solid state drive (SSD), a flash memory, etc. , and EEPROM (Electrically Erasable Programmable Read-Only Memory) The storage device 103 is used as a storage area for data and programs, a working area for the CPU 102, a buffer area for communication data, etc. The storage device 103 is used as a broadcast audio memory that stores text data for automatic broadcasting.

[0016] The CPU 102 communicates (transmits and receives information) with the train information management system 50. Railway vehicle operation information, vehicle information, and the like are received from the train information management system. The train information management system 50 is, for example, an integrated train ommunication network for evolvable railway operation system (INTEROS), a train information management system (TIMS), or a The vehicle information and operation information provided by the train information management system may include, for example, train type, train number or operation number, destination, current train location, next stop, etc. The vehicle information and operation information may also include information on the operating status of major equipment, door open / close status, in-car environment (temperature, humidity, occupancy rate), air conditioning settings, settings of automatic announcements and in-car guide display devices, operation support in emergencies and breakdowns, remote operation of equipment, etc.

[0017] CPU 102 acquires position information of the railway vehicle measured using a GPS, or the kilometre distance of the railway vehicle calculated using an encoder or the like, and reads out text data for automatic announcements in Japanese corresponding to the position of the railway vehicle from storage device 103 according to the position of the railway vehicle, and supplies this to audio signal generation unit 104. CPU 102 also transmits the text data for automatic announcements to terminal 10 through communication with terminal 10.

[0018] The audio signal generation unit 104 includes, for example, a circuit that generates an automatic broadcast audio signal by voice synthesis using text data, and a digital-to-analog converter (D / A) that converts the automatic broadcast audio signal from a digital signal format to an analog signal format, and outputs an analog automatic broadcast audio signal (Japanese audio signal).

[0019] The terminal 10 is an on-board terminal mounted on a railway vehicle, and may be, for example, a tablet terminal or a personal computer (PC). However, the type of the terminal 10 is not limited to these. The terminal 10 may be a multilingual terminal in which Japanese text data is translated into one or more predetermined languages. The SW 105 outputs a multilingual voice signal in a language other than Japanese. Languages ​​other than Japanese include, but are not limited to, English, German, Chinese, Korean, etc. The SW 105 selects and outputs one of the Japanese voice signal and the multilingual voice signal in response to a control signal from the CPU 102 or the like.

[0020] The control amplifier 106 activates the output amplifier 107. The output amplifier 107 amplifies the analog audio signal output from the SW 105 and connects it to the speaker 108. The speaker 108 emits audio (broadcast audio) corresponding to the audio signal. The microphone 109 generates an audio signal by collecting the emitted broadcast audio. The audio signal is input to the terminal 10.

[0021] 2 is a diagram showing an example of the configuration of terminal 10. Terminal 10 includes an interface circuit (IF) 11 for voice input, a noise processing unit 12, a voice recognition unit 13, an analog-digital converter (A / D) 14, a communication unit 15, a text processing unit 16, a voice synthesis unit 17, and a screen display control unit 18. Terminal 10 further includes a CPU 20, a DSP (Digital Signal Processor) 21, a storage device 22, an input device 23, a display 24, and a D / A 19.

[0022] The IF 11 is used to input a voice signal from the microphone 109. The noise processing unit 12 is a noise filter circuit that removes noise caused by the running sound of a railway vehicle, wind noise, etc. The voice recognition unit 13 performs voice recognition on the voice signal output from the noise processing unit 12 according to instructions from the CPU 20, for example, to generate text data of the voice signal, and supplies the text data to the CPU 20. The A / D 14 performs analog-to-digital conversion of the voice signal output from the noise processing unit 12, and inputs the digital voice signal to the DSP 21.

[0023] The communication unit 15 is a circuit that controls communication with the CPU 102, and receives Japanese text data supplied from the CPU 102. The Japanese text data received by the communication unit 15 is sent to the multilingual translation unit 30 and the CPU 20.

[0024] The multilingual translation unit 30 is, for example, an external server outside the terminal 10, and outputs multilingual text data corresponding to the Japanese text data. The multilingual text data is text data in which the wording expressed in the Japanese text data is expressed in one or more languages ​​other than Japanese. Note that the processing by the multilingual translation unit 30 may be an internal processing of the terminal 10.

[0025] The text processing unit 16 selects the language required for text comparison. For example, when the external input is Japanese voice or text, only Japanese is selected. Also, when comparing a foreign language (for example, English) translated by the multilingual translation unit 30, the foreign language (in this case) is selected. In this embodiment, only multilingual texts translated from Japanese to multiple languages ​​are selected. The DSP 21 extracts text data in a language to be compared (e.g., English) from the multilingual text data. The voice synthesis unit 17 inputs a digital voice signal generated by voice synthesis of the multilingual text data to the DSP 21. The screen display control unit 18 controls the display of text based on the multilingual text data on a display device provided in the passenger compartment. The screen display control unit 18 also performs display and operation when selecting a sentence to be broadcast inside the railway vehicle. The screen display control unit 18 has a memory, and stores text data of several standard phrases in the memory, and can display the standard phrase text on the display device at an appropriate timing. For example, one of the standard phrase text data can be selected under the control of the CPU 20 and input to the voice synthesis unit 17.

[0026] The storage device 22 stores the programs executed by the CPU 20 and the The CPU 20, by executing a program, uses the text comparison 213 to compare the Japanese text data (first text data) input from the communication unit 15 with the Japanese text data (second text data) input from the voice recognition unit 13, and calculates a voice recognition rate.

[0027] In addition, the text comparison 213 can also compare text data (first text data) other than Japanese input from the text processing unit 16 with text data (second text data) in multiple languages ​​input from the speech recognition unit 13 to calculate a speech recognition rate.

[0028] Here, the speech recognition rate indicates the probability with which the speech recognition unit 13 outputs the second text data, which is the correct data, when the first text data is taken as the correct data. For example, when the speech recognition unit 13 outputs the correct data 9 times out of 10 times, the speech recognition rate is 90%. The speech recognition rate may also indicate the correct answer rate (matching rate) of the second text data when the first text data is taken as the correct data. For example, when the first text data consists of 10 sentences and the second text data shows 9 correct sentences, the speech recognition rate is 90%. Alternatively, when the first text data is one sentence consisting of 10 characters, words, or morphemes, and the second text data shows 9 correct characters, words, or morphemes (90% correct), the speech recognition rate is 90%.

[0029] The CPU 20 adjusts the waveform of the voice signal (automatic broadcast voice signal) generated by voice synthesis of the first text data based on the voice recognition rate. The CPU 20 adjusts the waveform by at least one of volume adjustment, frequency characteristic adjustment, compressor adjustment, and limiter adjustment. The number and combination of these adjustment methods can be set appropriately. Volume adjustment is a process of increasing the amplitude of the voice signal overall. Frequency characteristic adjustment is a process of changing the frequency characteristic so that the voice recognition rate increases. The compressor is a process of compressing the volume of a waveform (peak) that exceeds a predetermined threshold at a predetermined compression rate, and in the compressor adjustment, in addition to compressing the peak, a process of increasing the overall signal level (to approach 0 dB) may be performed. The limiter is a process of lowering the volume of a part of the waveform that exceeds 0 dB so that it does not exceed 0 dB, and in the limiter adjustment, in addition to lowering the volume of the peak, a process of increasing the overall signal level (to approach 0 dB) may be performed. The compressor adjustment and limiter adjustment can increase the sound pressure level. In addition, sound distortion can be avoided by adjusting the limiter.

[0030] The result of such waveform adjustment, i.e., the signal processing parameters for obtaining the adjusted waveform, are stored in the storage device 22 by the CPU 20, for example, and are set by the CPU 20 at an appropriate timing to the DSP 21 (for languages ​​other than Japanese) that corrects the voice signal. This makes it possible to obtain a desired voice signal (with improved voice recognition rate). In the case of Japanese, the adjusted parameter information is transmitted to the CPU 102 and stored in the storage device 103, and the CPU 102 sets it in the voice signal generating unit 104 at an appropriate timing.

[0031] The DSP 21 operates as an adaptive filter for the input audio signal. The DSP 21 performs signal processing 211 for the audio signal input from the audio synthesis unit 17. In the signal processing 211, at least one of the volume adjustment, frequency characteristic adjustment, compressor adjustment, and limiter adjustment described above is performed according to parameters set by the CPU 20, for example.

[0032] The DSP 21 also performs signal comparison 212, which compares the audio signal input from the A / D 14 with the audio signal that has been subjected to signal processing 211. The signal comparison 212 detects waveforms in which the audio signal from the signal processing 211 has missing, low, distorted, etc., in the broadcast audio signal. The signal generation parameters used by the DSP 21 are adjusted so that the specified portion approaches the signal from the signal processing 211. The audio signal that has passed through the signal processing 211 is converted into an analog signal by the D / A and input to the SW 105.

[0033] The display 24 is a display device capable of displaying information and data, and is capable of displaying the signal waveform of the voice signal based on the first and second text data. The input device 23 is a key, button, knob, etc., and is used for manually adjusting the voice signal (volume adjustment, frequency characteristic adjustment, compressor adjustment, limiter adjustment) to improve the voice recognition rate.

[0034] In addition, some or all of the processing performed by CPU 102 or CPU 20 may be performed by a processor other than a CPU, such as a DSP (Digital Signal Processor), a dedicated or general-purpose integrated circuit (hardware) such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field Programmable Gate Array), or a combination of software and hardware (such as a SoC (System-on-a-Chip)).

[0035] Figures 3 and 4 are flowcharts showing an example of processing performed by the CPU 20 of the terminal 20. The processing shown in Figures 3 and 4 is processing in an automatic broadcast test performed in a railway vehicle undergoing a test run on a track.

[0036] In step S001, the CPU 20 determines whether or not a pre-adjustment is necessary based on the position of the railway vehicle using GPS position information, kilometre information, etc. If it is determined that a pre-adjustment is necessary, the pre-adjustment is performed (step S002).

[0037] In step S003, an automatic announcement using the first text data is actually made. The automatic announcement is also picked up by the microphone 109, and the second text data is generated. Then, the CPU 20 measures (calculates) a voice recognition rate by comparing the first text data with the second text data.

[0038] In step S004, the CPU 20 compares the voice recognition rate with a threshold value previously prepared in the storage device 22 to determine whether the voice recognition rate is equal to or higher than the threshold value. The threshold value is, for example, 80%, but may be greater or smaller than this. If it is determined that the voice recognition rate is equal to or higher than the threshold value, the process ends. If not, the process proceeds to step S005.

[0039] In step S005, the CPU 20 judges whether or not volume adjustment is necessary. This judgment may be made autonomously by the CPU 20 with reference to parameters indicating the current waveform and volume, or may be made based on information indicating a response to an inquiry made to a user (operator) of the terminal 10 as to whether volume adjustment is necessary. If it is judged that volume adjustment is necessary, the process proceeds to step S006, and if not, the process proceeds to step S008.

[0040] In step S006, the CPU 20 adjusts the volume. The volume adjustment may be performed, for example, by the user operating the input device 23 while referring to the waveform and parameters of the audio signal displayed on the display 24, or may be a process in which the CPU 20 increases the volume by a predetermined amount according to a predetermined algorithm.

[0041] In step S007, when the volume reaches a "volume OK" state (for example, the volume increases by a predetermined amount), the volume adjustment is terminated. Then, the process proceeds to step S016, where waveform adjustment is performed by the signal comparison 212. The adjustment based on the comparison result may be performed manually by the user, or may be performed automatically using some algorithm (for example, a macro that automates the procedure performed by the user).

[0042] When step S016 is completed, the process returns to step S003, and the speech recognition rate is recalculated using the adjusted speech signal. Note that the volume adjustment in step S006 may be performed only once, or may be performed two or more times.

[0043] In step S008, the CPU 20 determines whether or not the frequency characteristics need to be adjusted. This determination may be made autonomously by the CPU 20 with reference to a waveform or parameters indicating the current frequency characteristics, or may be made based on information indicating a response to an inquiry made to the user of the terminal 10 as to whether or not the frequency characteristics need to be adjusted. If it is determined that the frequency characteristics need to be adjusted, the process proceeds to step S009, and if not, the process proceeds to step S011.

[0044] In step S009, the CPU 20 adjusts the frequency characteristics. The adjustment of the frequency characteristics may be performed, for example, by the user operating the input device 23 while referring to the waveform and parameters of the audio signal displayed on the display 24, or may be performed by the CPU 20 automatically increasing or decreasing the power of the frequency by a predetermined amount for a predetermined frequency range according to a predetermined algorithm.

[0045] In step S010, when the frequency characteristics reach a "frequency characteristics OK" state (for example, the power in a specific frequency region increases or decreases by a predetermined amount), the frequency characteristic adjustment is terminated. Then, the process proceeds to step S016, where waveform adjustment is performed by signal comparison 212. When step S016 ends, the process returns to step S003, where the speech recognition rate is recalculated using the adjusted speech signal. Note that the frequency adjustment in step S008 may be performed only once, or may be performed two or more times.

[0046] In step S011, the CPU 20 determines whether or not adjustment of the peak and overall levels is necessary. This determination may be made autonomously by the CPU 20 with reference to the current signal waveform, or may be made based on information indicating a response to an inquiry made to the user of the terminal 10 as to whether adjustment is necessary. If it is determined in step S011 that adjustment is necessary, the process proceeds to step S012, and if not, the process proceeds to step S014.

[0047] In step S012, the CPU 20 performs compressor adjustment. The compressor adjustment may be performed, for example, by the user operating the input device 23 while referring to the waveform and parameters of the audio signal displayed on the display 24, or may be a process in which the CPU 20 compresses the volume of the waveform portion (peak) exceeding a predetermined threshold at a predetermined compression rate according to a predetermined algorithm. Furthermore, the compressor adjustment may further include a process of increasing the level of the entire waveform after the peak is compressed so that it does not exceed 0 dB.

[0048] In step S013, when the compressor adjustment reaches the "compressor adjustment OK" state (for example, each peak is volume-compressed and the overall level of the waveform increases), the compressor adjustment is terminated. Then, the process proceeds to step S016, where waveform adjustment is performed by the signal comparison 212. When step S016 ends, the process returns to step S003, where the speech recognition rate is recalculated using the adjusted speech signal. Note that step S012 may be performed only once, or may be performed two or more times by changing, for example, the threshold and compression rate.

[0049] In step S014, the CPU 20 performs limiter adjustment. The limiter adjustment may be performed, for example, by the user operating the input device 23 while referring to the waveform and parameters of the audio signal displayed on the display 24, or the CPU 20 may perform a process of lowering the volume of the portion (peak) in the waveform that exceeds 0 dB in accordance with a predetermined algorithm. Furthermore, in the limiter adjustment, a process of raising the level of the entire waveform after the volume of the peak has been lowered so that it does not exceed 0 dB may be further performed. Note that the process of raising the overall level of the waveform may be performed by only one of the compressor and the limiter.

[0050] In step S015, when the limiter adjustment reaches the "limiter adjustment OK" state (for example, the volume of each peak is lowered so as not to exceed 0 dB and the overall level of the waveform increases), the limiter adjustment is terminated. Then, the process proceeds to step S016, where waveform adjustment is performed by signal comparison 212. When step S016 ends, the process returns to step S003, where the speech recognition rate is recalculated using the adjusted speech signal. Step S015 may be performed once, or two or more times.

[0051] The adjustment by signal comparison (step S016) is optional and can be omitted. As the text to be compared, text of a section where the broadcast sound becomes difficult to hear due to the running sound of the railway vehicle during a test run of the railway vehicle may be extracted. Furthermore, text of a section where the broadcast sound becomes difficult to hear due to the volume strength or intonation in Japanese or a language other than Japanese may be extracted. Furthermore, the order of the volume adjustment, frequency characteristic adjustment, compressor adjustment, and limiter adjustment can be changed as appropriate.

[0052] According to the automatic broadcasting system of the embodiment, an automatic broadcasting voice is generated by voice synthesis of the first text data used for the automatic broadcasting. Also, second text data is generated by voice recognition of a voice signal obtained by collecting the broadcasting voice based on the automatic broadcasting voice signal. Then, the CPU 20 as a control unit adjusts the signal waveform of the automatic broadcasting voice signal according to the voice recognition rate obtained by comparing the first text data with the second text data. That is, the terminal 20 operates as an automatic broadcasting sound adjustment device including the CPU 20 as a control unit.

[0053] The CPU 20 adjusts the audio waveform of the automatic broadcast audio signal whose audio recognition rate is lower than a threshold (at least one of volume adjustment, frequency characteristic adjustment, compressor adjustment, and limiter adjustment). This adjustment increases the audio recognition rate. The increase in the audio recognition rate makes the automatic broadcast audio signal easier to hear. This reduces cases where the broadcast audio becomes difficult to hear due to sounds other than the broadcast audio inside and outside the train during operation, or where the broadcast audio becomes difficult to hear due to intonation in the broadcast language. Furthermore, the CPU 20 makes adjustments based on the results of the signal comparison 212, thereby making it possible to obtain an audio signal that is even easier to hear. The configuration of the embodiment described above can be modified as appropriate without departing from the scope of the present invention. [Explanation of symbols]

[0054] 10. Terminal 13. Voice recognition unit 20 CPU 21...DSP 22...Storage device 23 Input device 24. Display 102 CPU 103...Storage device 104 Voice recognition unit

Claims

1. A control unit that adjusts a signal waveform of the automatic broadcast audio signal according to a voice recognition rate obtained by comparing first text data used for the automatic broadcast with second text data generated from an audio signal obtained by collecting broadcast audio based on the automatic broadcast audio signal generated from the first text data. An automatic broadcast sound adjustment device including:

2. The control unit performs at least one of volume adjustment, frequency adjustment, compressor adjustment, and limiter adjustment on the automatic broadcast audio signal.

2. The automatic broadcast sound control device according to claim 1.

3. The control unit adjusts the automatic broadcast audio signal based on a comparison result between a waveform of the audio signal and a waveform of the automatic broadcast audio signal.

3. An automatic broadcast sound control device according to claim 1 or 2.

4. The first text data is text data for announcements on board a railway vehicle.

4. An automatic broadcast sound control device according to claim 1.

5. The first text data is text data in a language other than Japanese.

5. An automatic broadcast sound control device according to claim 1.

6. Calculating a speech recognition rate by comparing first text data used in the automatic broadcast with second text data generated from an audio signal obtained by collecting a broadcast sound based on the automatic broadcast audio signal generated from the first text data; adjusting a signal waveform of the automatic broadcast audio signal according to the voice recognition rate; The automatic broadcast sound adjustment method is executed by an information processing device.

7. At least one of volume adjustment, frequency adjustment, compressor adjustment, and limiter adjustment is performed on the automatic broadcast audio signal. The automatic broadcast sound adjustment method according to claim 6.

8. The automatic broadcast audio signal is adjusted based on a comparison result between a waveform of the audio signal and a waveform of the automatic broadcast audio signal.

8. The automatic broadcast sound adjustment method according to claim 6 or 7.

9. Calculating a speech recognition rate by comparing first text data used in the automatic broadcast with second text data generated from an audio signal obtained by collecting a broadcast sound based on the automatic broadcast audio signal generated from the first text data; adjusting a signal waveform of the automatic broadcast audio signal according to the voice recognition rate; A program that causes a computer to execute the following.

10. 10. The program according to claim 9, further comprising causing the computer to execute at least one of volume adjustment, frequency adjustment, compressor adjustment, and limiter adjustment on the automatic broadcast audio signal.

11. 11. The program according to claim 9, which causes the computer to adjust the automatic broadcast audio signal based on a comparison result between a waveform of the audio signal and a waveform of the automatic broadcast audio signal.

Images