JP2012059203A - Specific audio recognition apparatus and specific audio recognition method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To determine whether or not an acquired peripheral sound is matched to an area specific sound in a specific geographical area even if area specific sound data (such as sounds of a siren) used in the geographical area are different for each geographical area such as a municipality and if matched, to notify such a state in a vehicle.SOLUTION: A specific sound recognition apparatus includes: a sound acquisition section for acquiring an inputted peripheral sound and outputting the sound as digital sound data; and a current location generation section for calculating and outputting the current location of the specific sound recognition apparatus itself. Furthermore, the specific sound recognition apparatus includes an area specific sound storage section for storing an area specific sound that is heard only in a predetermined geographic area; and a current location recognition specification section for outputting a specific geographical area to which the current location belongs. Moreover, the specific sound recognition apparatus includes a sound recognition result notification section which compares the area specific sound of the specific geographical area with digital sound data and, when it is determined that both match, notifies it.

Description

  The present invention relates to a specific voice recognition device and a specific voice recognition method for notifying that when a voice unique to a certain geographical area such as a siren sound of an emergency vehicle is detected.

  According to the Road Traffic Law in recent years, when an emergency car (fire-fighting car, emergency car, or other car specified by a government ordinance) approaches while blowing a siren or the like, the vehicle approaches the left side of the road. You have to give way. In addition, at or near the intersection, it is stipulated that the vehicle must stop at the left side of the road (or the right side if necessary), avoiding the intersection. Here, the emergency vehicle means a vehicle in operation for the emergency service as defined by a Cabinet Order. Hereinafter, the emergency vehicle is also referred to as an emergency vehicle.

  However, if the volume of audio equipment inside the vehicle is high in a private vehicle, etc., or if the noise inside or outside the vehicle is high, the siren sound of an emergency vehicle outside the vehicle may be missed. The driver may not be able to take appropriate actions.

  In view of this, there has been proposed one technique for preventing the siren sound of an emergency vehicle from being missed (see, for example, Patent Document 1).

  The following description is made in the above-described Patent Document 1 “Vehicle Sound Introducing Device”.

  In other words, a plurality of microphones for collecting outside sounds are attached to the upper surface, rear surface, and left and right outer surfaces of the vehicle. The vehicle exterior sound collected from each microphone is selected by a microcomputer and amplified by an amplifier. The amplified vehicle exterior sound is configured to be emitted into the vehicle interior from a speaker installed in the vehicle corresponding to the microphone mounting position. As a result, the driver can surely hear the sound outside the vehicle useful for the driver of the vehicle.

  Further, even when listening with the sound of a car stereo or the like being closed with the window closed, the driver is informed of the state of emergency vehicle approach (see, for example, Patent Document 2).

  The following description is made in Patent Document 2 “Emergency vehicle detection alarm device and position display device” described above.

  That is, the microphone signal is input to the emergency vehicle recognition device, and the emergency vehicle recognition device detects the siren sound of the emergency vehicle and outputs the signal to the display controller and the voice controller. When a signal is input from the emergency vehicle recognition device, the display controller generates a video signal of a warning screen from the built-in video signal generation device and displays the warning screen on the display. The voice controller generates an alarm announcement from the built-in voice synthesizer and drives the speaker.

Japanese Patent Laid-Open No. 08-002339 (pages 2 to 4, FIGS. 1 to 5) Japanese Patent Laid-Open No. 06-325291 (pages 4-6, FIGS. 1-5)

  In the methods described in Patent Documents 1 and 2 described above, a comparison is made as to whether the vehicle exterior sound collected by the microphone outside the vehicle matches the sound such as the emergency vehicle siren sound registered in the storage means in the vehicle in advance. I do. And if it corresponds, it recognizes that the emergency vehicle is approaching, and notifies a driver | operator with the screen and speaker in a vehicle. In Patent Document 2, sounds such as siren sounds of emergency vehicles registered in the storage means in the vehicle are stored separately for fire-fighting vehicles, emergency vehicles, police-related vehicles (such as police cars), and the like.

  However, although the voice data such as the siren sound of the emergency vehicle may vary depending on the local government or the like, the above-described Patent Documents 1 and 2 have a problem that this is not taken into consideration. .

  That is, in the case of a fire-fighting vehicle using a motor siren, the frequency of the siren sound is continuously changed during a predetermined time. And the standard of the frequency to be used is often about 250 to 850 Hz. Moreover, the frequency of the siren when it rings when passing an intersection etc. is about 900 Hz. Note that the motor siren is a siren that makes a sound by rotating its wings with a motor. However, in recent years, the sound of the motor siren may be replaced with an electronic sound.

  And since only the standard is shown about the grade of the change of the frequency of a siren sound, the duration of a sound, etc., it becomes possible to use a different siren sound for each local government.

The present invention has been made to solve the above-described problems. Therefore, the object of the present invention is to
Even if the area-specific voice data (siren sounds, etc.) used in the geographical area is different for each geographical area such as a local government, the acquired surrounding voice is the area of the specific geographical area. To provide a specific speech recognition device and a specific speech recognition method that make it possible to determine whether or not the voice matches the specific voice and, if the voice matches, notify the vehicle of that fact. .

In order to achieve the above object, a specific speech recognition apparatus according to the present invention acquires a peripheral sound to be input in a unit of a predetermined time length and outputs it as digital sound data, and a predetermined signal to be input Based on the current position generation unit that calculates and outputs its current position, and stores area-specific speech that is heard only in a predetermined geographical area in association with the geographical area in units of the predetermined time length or more. A region-specific voice storage unit; a current position recognition specifying unit that outputs information indicating a specific geographical region that is the geographical region to which the current position belongs, input from the current position generation unit; and the input specific geography If the region specific sound extracted from the region specific sound storage unit for each target region is compared with the digital sound data input from the sound acquisition unit, and it is determined that both match, And it outputs a notification signal for notifying includes a speech recognition result notification unit.

  Also, the specific speech recognition method of the present invention acquires peripheral speech as analog speech data in a unit of a certain time length, converts the analog speech data into digital speech data by A / D conversion, and a plurality of them at regular time intervals. Receiving the radio wave transmitted from the GPS satellite, calculating the current position based on the received radio wave information, identifying a plurality of local governments in the vicinity of the current position based on the current position, and using the identified plurality of local governments A plurality of emergency vehicle voice data is extracted, and if the digital voice data matches any of a plurality of emergency vehicle voice data of a plurality of local governments in the vicinity of the current position. Is output and notified by voice, message data, or image data.

  According to the present invention, even if the region-specific voice (such as a siren sounded by an emergency vehicle) used by the geographical region is different for each geographical region (such as a local government), the acquired surroundings Can be determined to match the region-specific speech of a particular geographic region. And when it corresponds, it becomes possible to alert | report to that effect in a vehicle.

It is a block diagram which shows 1st Embodiment of the specific speech recognition apparatus of this invention. It is a block diagram which shows 2nd Embodiment of the specific speech recognition apparatus of this invention. It is a sequence diagram explaining operation | movement of 2nd Embodiment of a specific speech recognition apparatus. It is a block diagram which shows 3rd Embodiment of the specific speech recognition apparatus of this invention. It is a figure which shows an example of the waveform of digital audio data. It is a figure which shows an example of the waveform data of the power spectrum for every frequency. It is a figure which shows an example of the data format of sonagram pattern data. It is a figure which shows an example of the example of a display of the pattern data of a sonogram. It is a sequence diagram explaining operation | movement of 3rd Embodiment of a specific speech recognition apparatus. It is a block diagram which shows 4th Embodiment of the specific speech recognition apparatus of this invention. It is a sequence diagram explaining operation | movement of 3rd Embodiment of a specific speech recognition apparatus.

Next, embodiments of the present invention will be described with reference to the drawings.
[First Embodiment]

  FIG. 1 is a block diagram showing a first embodiment of the specific speech recognition apparatus of the present invention.

  A specific speech recognition apparatus 100 shown in FIG. 1 includes a speech acquisition unit 2, a current position generation unit 4, and a region specific speech storage unit 6. The specific speech recognition apparatus 100 includes a current position recognition specification unit 3-2 and a speech recognition result notification unit 3-6.

  The voice acquisition unit 2 acquires the voice around the vehicle on which the specific voice recognition device 100 is mounted in a unit of a certain time length, and outputs it as digital voice data to the voice recognition result notification unit 3-6.

  The current position generation unit 4 calculates the current position of the specific speech recognition apparatus 100 based on the radio signal received from the GPS satellite, and outputs the current position to the current position recognition specifying unit 3-2.

  The area specific sound storage unit 6 stores, as area specific sounds, siren sounds generated by emergency vehicles that are heard only in a specific geographical area such as a local government. The area-specific speech is stored in association with the geographical area.

  The current position recognition specifying unit 3-2 specifies where the current position input from the current position generating unit 4 is located in the geographical area, and the voice recognition result is used as information on the specific geographical area. It outputs to the alerting | reporting part 3-6.

  The speech recognition result notifying unit 3-6 searches the region specific speech storage unit 6 based on the specific geographical region input from the current position recognition specifying unit 3-2, and extracts the region specific speech corresponding to the specific geographical region. To do. Then, the extracted region specific voice is compared with the digital voice data input from the voice acquisition unit 2. As a result of the comparison, if both match, it is determined that the emergency vehicle is approaching the vehicle on which the specific speech recognition device 100 is mounted. And the alerting | reporting signal which alert | reports that the emergency vehicle is approaching is output in the vehicle carrying the specific speech recognition apparatus 100. FIG.

  As described above, the specific speech recognition apparatus 100 according to the present embodiment includes the speech acquisition unit 2 that acquires input peripheral speech in units of a predetermined time length and outputs the digital speech data.

  Further, it includes a current position generator 4 that calculates and outputs its current position based on a predetermined input signal.

  Furthermore, a region-specific speech storage unit 6 is provided that stores region-specific speech heard only in a predetermined geographic region in association with the geographic region in units of the predetermined time length or more.

  Moreover, the present position recognition specific | specification part 3-2 which outputs the information which shows the specific geographical area which is the said geographical area to which the said present position belongs input from the said present position production | generation part 4 is included.

  Furthermore, the speech recognition result notifying unit 3- compares the region specific speech extracted from the region specific speech storage unit 6 with respect to the input specific geographical region and the digital speech data input from the speech acquisition unit 2. 6 is included. In addition, when it is determined that both the region specific voice and the digital voice data match, the voice recognition result notification unit 3-6 outputs a notification signal to notify that fact.

Therefore, according to the present embodiment, even if the region specific sound (such as a siren sounded by an emergency vehicle) used by the geographical region is different for each geographical region (such as a local government), It can be determined whether the acquired surrounding voice matches the area specific voice of a specific geographical area. And when it corresponds, it becomes possible to alert | report to that effect in a vehicle.
[Second Embodiment]
FIG. 2 is a block diagram showing a second embodiment of the specific speech recognition apparatus of the present invention.

  A specific speech recognition apparatus 100 shown in FIG. 2 includes a microphone 10 corresponding to the speech acquisition unit 2 in FIG. 1 and a speech data extraction unit 20. Moreover, the control part 30 is included.

  Furthermore, a GPS (Global Positioning System) receiving unit 40 corresponding to the current position generating unit 4 of FIG. 1 is included.

  Moreover, the emergency vehicle audio | voice data storage part 60 connected to the control part 30 corresponded to the area | region specific audio | voice storage part 6 of FIG. 1 is included.

  Furthermore, the specific speech recognition apparatus 100 includes a map data storage unit 50 connected to the control unit 30 corresponding to the current position recognition specifying unit 3-2 in FIG. 1, a current position specifying unit 32 included in the control unit 30, Is included.

  Moreover, the emergency vehicle audio | voice data selection part 34 and the warning sound recognition part 36 equivalent to the audio | voice recognition result alerting | reporting part 3-6 of FIG. Furthermore, the specific speech recognition apparatus 100 includes an alarm output unit 70 connected to the control unit 30.

  The microphone 10 is installed outside the vehicle on which the specific speech recognition device 100 is mounted, acquires sound around the microphone 10, and sends it to the sound data extraction unit 20.

  The voice data extraction unit 20 converts the analog voice data sent from the microphone 10 into digital voice data and sends it to the control unit 30.

  The GPS receiver 40 receives radio waves from GPS satellites, calculates current position data of the specific speech recognition apparatus 100 as, for example, latitude and longitude data, and sends the data to the controller 30.

  The map data storage unit 50 stores, for example, map data for the whole of Japan together with latitude and longitude data, and, in addition, any local government (□□ prefecture, XX city, It also stores information about whether it belongs to a town.

  The emergency vehicle voice data storage unit 60 stores emergency vehicle voice data such as a siren sound generated by an emergency vehicle of the local government for each local government. The types of emergency vehicles are classified into, for example, fire-fighting vehicles, emergency vehicles, patrol cars, and the like, and the emergency vehicle audio data storage unit 60 stores emergency vehicle audio data for each type of emergency vehicle. Note that the emergency vehicle audio data in the present embodiment is stored in the form of a sonagram (voiceprint) that expresses audio waveform information in three dimensions of time, frequency, and intensity. The details of the sonagram will be described later in the third embodiment.

  The current position specifying unit 32 of the control unit 30 receives the current position data of the specific speech recognition apparatus 100 from the GPS receiving unit 40. Then, the current position specifying unit 32 specifies which municipality the received current position data corresponds to by referring to the map data storage unit 50. Note that the current position specifying unit 32 also specifies other two or three local governments in the vicinity of the specified local government. This is because, when the specific speech recognition apparatus 100 is located around a boundary of a certain local government, in addition to the emergency vehicle of the local government indicated by the current position data, the emergency vehicle of the emergency vehicle of another local government in the vicinity of the local government This is because it is possible that audio data is acquired. The current position specifying unit 32 sends information on a plurality of local governments specified by the current position specifying unit 32 to the emergency vehicle voice data selecting unit 34.

  The emergency vehicle voice data selection unit 34 of the control unit 30 based on the information of the plurality of local governments received from the current position specifying unit 32, the emergency vehicle voice data storage unit 60 stores the emergency vehicle voice data of the emergency vehicles of the plurality of local governments. And is sent to the alarm sound recognition unit 36.

  The alarm sound recognition unit 36 of the control unit 30 first converts the digital audio data received from the audio data extraction unit 20 into three-dimensional data in the form of a sonagram (hereinafter referred to as reception sonagram data). Next, the received sonogram data is compared with emergency vehicle voice data (in the form of a sonogram) of emergency vehicles of a plurality of local governments received from the emergency vehicle voice data selection unit 34. If any of the emergency vehicle audio data of a plurality of municipalities received from the emergency vehicle audio data selection unit 34 matches the received sonargram data, it is determined that the received sonagram data is audio data that the emergency vehicle is blowing. . If it is determined that the emergency vehicle audio data is the emergency vehicle, the warning sound recognition unit 36 sends a notification to the warning output unit 70 that the emergency vehicle is approaching.

  The alarm output unit 70 outputs and notifies that the emergency vehicle is approaching the specific voice recognition device 100 by voice, message data, or image data.

  Next, the operation of the present embodiment will be described with reference to FIG.

  FIG. 3 is a sequence diagram for explaining the operation of the present embodiment.

  In FIG. 3, the microphone 10 of the specific speech recognition apparatus 100 acquires the sound around the microphone 10 in a unit of a predetermined time length and sends it to the sound data extraction unit 20 (step S <b> 1 in FIG. 3). After that, step S1 is repeated to acquire the next time length voice.

  The audio data extraction unit 20 converts analog audio data sent from the microphone 10 into digital audio data (A / D conversion: Analog to Digital Conversion) (step S2). Then, the converted digital audio data is sent to the control unit 30.

  On the other hand, the GPS receiver 40 of the specific speech recognition apparatus 100 receives radio waves transmitted by a plurality of GPS satellites at regular time intervals, and calculates the current position of the specific speech recognition apparatus 100 based on the received radio wave information ( Step S3). Then, the calculated current position information is sent to the control unit 30. Thereafter, step S3 is repeated to receive radio waves from GPS satellites at the next time interval.

  The control unit 30 that has received the current position from the GPS receiving unit 40 specifies which local government the received current position corresponds to by referring to the map data storage unit 50 shown in FIG. Furthermore, the control unit 30 also specifies other 2-3 local governments in the vicinity of the specified local government (step S4).

  Next, the control unit 30 extracts emergency vehicle sound data used by the plurality of specified local governments from the emergency vehicle sound data storage unit 60 shown in FIG. 2 (step S5). The number of emergency vehicle voice data extracted here is the number of types of emergency vehicles in a plurality of local governments, and the emergency vehicle voice data is three-dimensional data in the form of a sonogram.

On the other hand, the control unit 30 first receives the digital audio data received from the audio data extraction unit 20,
It is converted into three-dimensional data in the form of sonagram (referred to as received sonagram data).

  Next, the control unit 30 compares the received sonargram data with any of a plurality of emergency vehicle voice data (also represented by sonagrams) of a plurality of local governments extracted in step S5. (Step S6). Then, the control unit 30 determines whether the received sonogram data matches any of the plurality of emergency vehicle voice data of the plurality of local governments extracted in step S5 (step S7). The coincidence determination may be performed by pattern matching between data represented by a sonogram.

  Here, when it is determined that the received sonagram data matches any of a plurality of emergency vehicle voice data of a plurality of municipalities extracted in step S5 (“match” in step S7), the control unit 30 approaches the emergency vehicle. Judge that there is. And it notifies the alarm output part 70 of the specific speech recognition apparatus 100 that the emergency vehicle is approaching. The warning output unit 70 outputs and notifies that an emergency vehicle is approaching by voice, message data, or image data (step S8). When the user of the specific speech recognition apparatus 100 notices this notification and performs a notification stop operation (for example, pressing a notification stop button (not shown)), the alarm output unit 70 stops the alarm notification. Return to the initial state.

  In step S7, when the received sonagram data does not match any of the plurality of emergency vehicle audio data of the plurality of local governments extracted in step S5 (“mismatch” in step S7), the control unit 30 does not do anything, Return to the initial state.

  The operation of the second embodiment of the present invention has been described above.

  As described above, the specific speech recognition apparatus 100 according to the present embodiment performs A / D conversion on the analog audio data acquired by the microphone 10 that acquires peripheral audio in units of a certain time length and digital audio. And an audio data extraction unit 20 to be sent to the control unit 30 as data.

  And a GPS receiving unit 40 that receives radio waves transmitted by a plurality of GPS satellites at regular time intervals, calculates a current position based on the received radio wave information, and transmits the current position to the control unit 30.

  Furthermore, it includes an emergency vehicle voice data storage unit 60 that is connected to the control unit 30 and stores emergency vehicle voice data such as a siren sound generated by an emergency vehicle of the local government for each local government.

  Moreover, the map data storage part 50 connected to the said control part and memorize | stores map data is included.

  Furthermore, a current position specifying unit 32 that is included in the control unit 30 and specifies a plurality of local governments in the vicinity of the current position with reference to the map data storage unit 50 based on the current position received from the GPS receiving unit 40, Is included.

  The emergency vehicle voice data selection unit 34 that extracts emergency vehicle voice data included in the control unit 30 and used by a plurality of local governments specified by the current position specifying unit from the emergency vehicle voice data storage unit 60; , Including.

  Further, an alarm sound recognition unit that determines whether the digital sound data received from the sound data extraction unit 20 matches any of a plurality of emergency vehicle sound data of a plurality of local governments extracted by the emergency vehicle sound data selection unit 34. 36. When the warning sound recognition unit 36 determines that they match, the warning sound recognition unit 36 notifies the warning output unit 70 connected to the control unit 30 that an emergency vehicle is approaching.

  The alarm output unit 70 is configured to output and notify that an emergency vehicle is approaching by voice, message data, or image data.

Therefore, according to the present embodiment, even when the emergency vehicle voice data (siren sound etc.) used by the local government is different for each local government, the voice data outside the vehicle is blown by the emergency vehicle. It can be determined whether or not it matches the emergency vehicle voice data. And when it corresponds, it becomes possible to alert | report in the vehicle that the emergency vehicle is approaching.
[Third embodiment]
Next, a third embodiment according to the present invention that further embodies the second embodiment shown in FIG. 2 will be described.

  FIG. 4 is a block diagram showing a third embodiment of the specific speech recognition apparatus of the present invention. FIG. 4 is a block diagram for explaining the specific speech recognition apparatus shown in FIG. 2 in more detail. Therefore, in FIG. 4, the same reference numerals or symbols are assigned to the components corresponding to those in FIG. 2, and the description thereof is omitted as much as possible.

  The specific speech recognition apparatus 100 shown in FIG. 4 includes a microphone 10, a speech data extraction unit 20, a control unit 30, and a GPS reception unit 40 as in FIG. 2. The specific voice recognition device 100 includes a map data storage unit 50, an emergency vehicle voice data storage unit 60, and an alarm output unit 70 connected to the control unit 30.

  The microphone 10 acquires sound around the microphone 10 and sends it to the sound data extraction unit 20. The microphone 10 is assumed to be installed outside the vehicle on which the specific speech recognition apparatus 100 is mounted and at a place where external voice data can be easily obtained.

  The audio data extracting unit 20 includes an amplifier 22, a band pass filter (Band Pass Filter) 24, and an A / D converter (Analog to Digital Converter) 26.

  The amplifier 22 amplifies the analog audio data sent from the microphone 10 and sends it to the bandpass filter 24.

The band-pass filter 24 is a filter circuit that normally passes only frequencies in a necessary range and does not pass other frequencies. The band-pass filter 24 of the present embodiment passes audio data only in the frequency range (for example, 250 to 900 Hz) of the siren sound that the emergency vehicle blows.
The audio data is sent to the A / D converter 26.

  The A / D converter 26 converts the analog audio data received from the bandpass filter 24 into digital audio data, and sends the digital audio data to the control unit 30.

  The control unit 30 includes a CPU (Central Processing Unit) 37 that performs arithmetic processing of the specific speech recognition apparatus 100. Further, the control unit 30 includes a ROM (Read Only Memory) 38 for storing a program for operating the CPU 37, emergency vehicle approach warning image data, emergency vehicle approach warning message data, and the like. Programs that are stored in the ROM 38 and operate the CPU 37 include programs that execute the functions of the current position specifying unit 32, the emergency vehicle voice data selection unit 34, and the alarm sound recognition unit 36 described in the second embodiment. ing.

  Further, the control unit 30 includes a RAM (Random Access Memory) 39 that stores various data as a work area of the CPU 37.

  The GPS receiver 40 calculates the current position data of the specific speech recognition apparatus 100 as, for example, latitude and longitude data, based on the GPS antenna 42 that receives radio waves from GPS satellites and the radio waves received from the GPS antenna 42. A GPS receiver 44 is included. The GPS receiver 44 sends the calculated current position data to the control unit 30.

  The map data storage unit 50 includes, for example, map data 52 that stores map data of the whole of Japan together with latitude and longitude data. The map data 52 further stores information on which local government in Japan (□□ prefecture, XX city, △ Δ town, etc.) the latitude and longitude points belong to.

  The emergency vehicle audio data storage unit 60 includes emergency vehicle audio data 62 for storing emergency vehicle audio data such as sirens sounded by an emergency vehicle of the local government for each local government. The emergency vehicle voice data 62 stores the respective voice data as the types of emergency vehicles, for example, for fire fighting cars, emergency cars, patrol cars, and the like. In addition, the audio | voice data of the emergency vehicle in this embodiment shall be memorize | stored in the form of the sonagram which expressed the information of the audio | voice waveform in three dimensions of time, a frequency, and an intensity | strength.

  Here, with reference to FIG. 5, FIG. 6, FIG. 7, and FIG. 8, the sonagram used in this embodiment will be described.

  FIG. 5 is a diagram showing an example of the waveform of the digital audio data sent from the A / D converter 26 of the audio data extraction unit 20 to the control unit 30. The CPU 37 of the control unit 30 stores this digital audio data in the RAM 39. The digital audio data shown in FIG. 5 is divided at intervals of a fixed time, and each fixed time is expressed as “T1” to “Tn”. Then, the CPU 37 converts this digital audio data into data in the form of a sonogram in the following procedure.

  That is, the CPU 37 Fourier transforms the digital audio data (FIG. 5) stored in the RAM 39 for each predetermined time “T1” to “Tn” to obtain waveform data of the power spectrum for each frequency. FIG. 6 is a diagram illustrating an example of waveform data of a power spectrum for each frequency in a certain time “T1”, for example. In the case of the present embodiment, the frequency range of the waveform data shown in FIG. 6 is, for example, 250 to 900 Hz (because it is filtered by the bandpass filter 24). Note that if FFT (Fast Fourier Transform) is used for Fourier transform, high-speed processing is possible, which is convenient.

  Next, with respect to the waveform data (FIG. 6) of the power spectrum for each frequency, the frequency exceeding the threshold value 200 shown in the middle of FIG. 6 is obtained, and the correspondence data between the frequency and time is obtained. Remember me. Here, FIG. 6 shows the waveform data of the power spectrum for each frequency for a certain time “T1” as described above. The threshold value 200 in FIG. 6 is a value for classifying the power, that is, the intensity of each frequency, into two values, the upper side (higher intensity) or the lower side (lower intensity) of the threshold 200. . If the frequency above the threshold 200 is represented by “1” and the frequency below the threshold 200 is represented by “0”, as shown in the bottom row (“T1” time row 601) of FIG. It can be shown as data of “f1” to “fm” (641 to 64m columns in FIG. 7).

  Similarly to FIG. 6, the waveform data of the power spectrum for each frequency is obtained for the fixed times “T2” to “Tn”. Similarly to the above, if “1” and “0” represent the upper side or the lower side of the “threshold value”, the 602 row (“T2” time row) to the 60n row (“Tn” time) in FIG. The data of the frequencies “f1” to “fm” (columns 641 to 64m in FIG. 7) as shown in FIG.

That is, in FIG. 7, the digital audio data shown in FIG. 5 is converted into time (“T1” to “Tn”).
, Frequency (“f1” to “fm”) and intensity (“1” or “0”). Accordingly, FIG. 7 shows an example of the data format of the sonagram pattern data.

  Note that the threshold 200 shown in FIG. 6 is configured so that the value can be changed. With this configuration, the value expressed as the intensity for each frequency can be changed according to the surrounding situation (noise level, etc.). In the above description, it has been described that only one threshold 200 is set and the intensity is classified into binary. However, a plurality of thresholds are set so that the intensity can be classified into a plurality of intensity such as 4-value and 8-value. You can make it possible. By configuring in this way, the intensity shown in the sonagram pattern data (first example) 600 of FIG. 7 is changed to not only binary but also quaternary or octal, so that more detailed sonagram data can be obtained. Is possible.

  FIG. 8 is a diagram illustrating an example of display of sonagram pattern data, and is referred to as sonagram pattern data (second example) 700. That is, the “1” portion shown in FIG. 7 is represented as “shaded portion” in FIG. 8, and the “0” portion in FIG. 7 is represented as “blank portion” in FIG. As the sonagram pattern data, either FIG. 7 or FIG. 8 may be used. However, depending on the pattern matching method to be described later, which of FIGS. 7 and 8 is suitable may be different. Therefore, it is preferable that the pattern matching can be performed in either of FIGS.

  Returning to FIG. 4, the alarm output unit 70 of the specific speech recognition apparatus 100 includes a display 72 and a speaker 74. Then, message data or image data output from the CPU 37 of the control unit 30 and reporting that the emergency vehicle is approaching the specific voice recognition apparatus 100 is output and notified. Further, the fact that the emergency vehicle is approaching the specific voice recognition device 100 is output by voice from the speaker 74 and notified.

  Next, the operation of the third embodiment will be described with reference to FIG.

  In the third embodiment of the present invention, the functional configuration of the specific speech recognition apparatus 100 shown in FIG. 2 is changed to a hardware representation as shown in FIG. Accordingly, the operation shown in FIG. 9 is almost the same as the operation shown in FIG. 3. In FIG. 9, the operation shown in FIG. 3 is denoted by the same reference numeral or symbol, and the description thereof is omitted as much as possible. .

  FIG. 9 is a sequence diagram for explaining the operation of the third embodiment.

  In FIG. 9, the microphone 10 of the specific speech recognition apparatus 100 acquires the sound around the microphone 10 in a unit of a certain time length and sends it to the sound data extraction unit 20 (step S1 in FIG. 9). After that, step S1 is repeated to acquire the next time length voice.

  The audio data extraction unit 20 amplifies the analog audio data sent from the microphone 10 with the amplifier 22 and sends it to the bandpass filter 24. The band-pass filter 24 passes the audio data only in the frequency range (for example, 250 to 900 Hz) of the siren sound that the emergency vehicle blows and sends it to the A / D converter 26. The A / D converter 26 converts the analog audio data received from the bandpass filter 24 into digital audio data (step S2-1). Then, the converted digital audio data is sent to the control unit 30.

  On the other hand, the GPS receiving unit 40 of the specific speech recognition apparatus 100 receives the radio waves transmitted from a plurality of GPS satellites by the GPS antenna 42 at regular time intervals. Then, based on the radio wave information received by the GPS antenna 42, the GPS receiver 44 calculates the current position of the specific speech recognition device 100 (step S3-1). Then, the calculated current position information is sent to the control unit 30. Thereafter, step S3-1 is repeated to receive radio waves from the GPS satellite at the next time interval.

  The CPU 37 of the control unit 30 that has received the current position from the GPS receiver 44 of the GPS receiving unit 40 causes the program of the current position specifying unit (current position specifying unit 32 in FIG. 2) to be executed. Then, the CPU 37 specifies which municipality the received current position corresponds to by referring to the map data 52 of the map data storage unit 50 shown in FIG. Furthermore, the CPU 37 also specifies other 2-3 local governments in the vicinity of the specified local government (step S4-1).

  Next, the CPU 37 of the control unit 30 causes the emergency vehicle audio data selection unit (emergency vehicle audio data selection unit 34 in FIG. 2) to execute a program. And the emergency vehicle audio | voice data which the some local government specified by step S4-1 are using are extracted from the emergency vehicle audio | voice data 62 of the emergency vehicle audio | voice data storage part 60 shown in FIG. 4 (step S5-1). The number of emergency vehicle voice data extracted here is the number of types of emergency vehicles in a plurality of local governments, and the emergency vehicle voice data is three-dimensional data in the form of a sonogram.

  On the other hand, when the CPU 37 of the control unit 30 receives the digital audio data from the A / D converter 26 of the audio data extraction unit 20, the CPU 37 executes the program of the alarm sound recognition unit (the alarm sound recognition unit 36 in FIG. 2). The CPU 37 first converts the digital audio data into three-dimensional data in the form of sonagram (referred to as received sonargram data) and stores it in the RAM 39. Next, the CPU 37 compares the received sonogram data with any of the plurality of emergency vehicle voice data of the plurality of local governments extracted at step S5-1 (step S6-1). Then, the CPU 37 determines whether the received sonargram data matches any of the plurality of emergency vehicle sound data of the plurality of local governments extracted in step S5-1 (step S7-1). The coincidence is determined by pattern matching between the data represented by the sonagram.

  When it is determined in step S7-1 in FIG. 9 that the received sonogram data matches any of a plurality of emergency vehicle voice data of a plurality of local governments extracted in step S5-1 (“match” in step S7-1). ), The CPU 37 of the control unit 30 determines that the emergency vehicle is approaching. Then, emergency vehicle approach warning image data indicating that an emergency vehicle is approaching, emergency vehicle approach warning message data, and the like are read from the ROM 38 and notified to the alarm output unit 70 of the specific speech recognition apparatus 100. The alarm output unit 70 notifies the display 72 that the emergency vehicle is approaching by using image data or message data. Alternatively, the fact that the emergency vehicle is approaching is output by voice as message data from the speaker 74 and notified (step S8-1). When the user of the specific speech recognition apparatus 100 notices this notification and performs a notification stop operation (for example, pressing a notification stop button (not shown)), the alarm output unit 70 stops the alarm notification. Return to the initial state.

  In step S7-1, when the received sonagram data does not match any of the plurality of emergency vehicle audio data of the plurality of local governments extracted in step S5-1 (“mismatch” in step S7-1), the CPU 37 of the control unit 30 Does nothing and returns to the original initial state.

  The operation of the third embodiment of the present invention has been described above.

  As described above, in the third embodiment of the present invention, the functional configuration of the specific speech recognition apparatus of the second embodiment shown in FIG. 2 is detailed as a hardware configuration as shown in FIG. It is. Therefore, in the following, description of the overall function of the third embodiment will be omitted, and only the parts specific to the third embodiment will be described.

  As described above, the voice data extraction unit 20 of the specific voice recognition apparatus 100 of this embodiment includes the amplifier 22, the bandpass filter 24, and the A / D converter 26. The band-pass filter 24 passes audio data only in the frequency range of the siren sound that the emergency vehicle blows.

  In addition, the map data storage unit 50 connected to the control unit 30 of the specific speech recognition apparatus 100 also expresses information on which municipality the latitude and longitude points belong together with the latitude and longitude data. Data 52 is included.

  Furthermore, the emergency vehicle voice data storage unit 60 connected to the control unit 30 of the specific voice recognition apparatus 100 stores emergency vehicle voice data in the form of a sonagram for each local government, which stores emergency vehicle voice data generated by the local emergency vehicle. Data 62 is included.

  In addition, the CPU 37 included in the control unit 30 executes a current position specifying unit program, an emergency vehicle voice data selection unit program, an alarm sound recognition unit program, and the like stored in the ROM 38, and performs the following operations.

  That is, when the CPU 37 receives the current position from the GPS receiver 44 of the GPS receiving unit 40, the CPU 37 refers to the map data 52 of the map data storage unit 50 as to which local government the received current position corresponds to. To identify. In addition, the CPU 37 also specifies other 2-3 local governments in the vicinity of the specified local government.

  Next, the CPU 37 extracts emergency vehicle sound data used by the specified plurality of local governments from the emergency vehicle sound data 62 of the emergency vehicle sound data storage unit 60. The number of emergency vehicle voice data extracted here is the number of types of emergency vehicles in a plurality of local governments, and the emergency vehicle voice data is three-dimensional data in the form of a sonogram.

  On the other hand, when the CPU 37 receives the digital audio data from the A / D converter 26 of the audio data extraction unit 20, the CPU 37 first converts the digital audio data into three-dimensional data in the form of sonagram (referred to as received sonagram data). Next, the CPU 37 determines whether or not the received sonargram data matches any of a plurality of emergency vehicle sound data of a plurality of local governments in the vicinity of the current position (these are also data represented by sonagrams). The coincidence is determined by pattern matching between the data represented by the sonagram.

  When the CPU 37 determines that the received sonargram data matches any of a plurality of emergency vehicle audio data of a plurality of local governments in the vicinity of the current position, an alarm output connected to the control unit 30 that the emergency vehicle is approaching Notification to the unit 70.

  When the CPU 37 receives a notification that the emergency vehicle is approaching, the alarm output unit 70 outputs the fact that the emergency vehicle is approaching from the display 72 or the speaker 74 by image data, message data, or voice. And informs you.

  Therefore, according to the present embodiment, even when the emergency vehicle voice data (siren sound etc.) used by the local government is different for each local government, the voice data outside the vehicle is blown by the emergency vehicle. It can be determined whether or not it matches the emergency vehicle voice data. And when it corresponds, it becomes possible to alert | report in the vehicle that the emergency vehicle is approaching.

  Further, the band-pass filter 24 can extract only voice data in a frequency range where the emergency vehicle blows, and the received voice data can be efficiently converted into a sonogram.

Further, the emergency vehicle audio data 62 stores a large number of audio data that sounds from emergency vehicles of all local governments. Only data can be extracted. Therefore, it is possible to greatly reduce the number of times of pattern matching of the sonagram data type.
[Fourth Embodiment]
Next, a description will be given of a fourth embodiment of the present invention in which some of the components of the third embodiment shown in FIG. 4 are replaced with a car navigation system.

  FIG. 10 is a block diagram showing a fourth embodiment of the specific speech recognition apparatus of the present invention. 10 that correspond to the components in FIG. 4 are denoted by the same reference numerals or symbols, and the description thereof is omitted as much as possible.

  A specific voice recognition device 100-1 shown in FIG. 10 includes a microphone 10, a voice data extraction unit 20, a control unit 30, and an emergency vehicle voice data storage unit 60 connected to the control unit 30.

  The microphone 10 is the same as in FIG. 4, acquires sound around the microphone 10, and sends it to the sound data extraction unit 20. The microphone 10 is assumed to be installed outside the vehicle on which the specific speech recognition device 100-1 is mounted and at a place where external speech data can be easily obtained.

  The audio data extraction unit 20 is the same as that in FIG. 4, and includes an amplifier 22, a band pass filter 24, and an A / D converter 26. And since the function of each component is the same as that shown in FIG. 4, further explanation is omitted.

  The control unit 30 includes a CPU 37-1 that performs arithmetic processing of the specific speech recognition apparatus 100-1. The control unit 30 includes a ROM 38-1 for storing a program for operating the CPU 37-1. The programs that are stored in the ROM 38-1 and operate the CPU 37-1 include programs that execute the functions of the emergency vehicle voice data selection unit 34 and the alarm sound recognition unit 36 described in the second embodiment.

  Furthermore, the control unit 30 includes a RAM 39-1 that stores various data as a work area of the CPU 37-1.

  Moreover, the emergency vehicle audio | voice data storage part 60 connected to the control part 30 is the same as that of FIG. 4, and the emergency vehicle which memorize | stores emergency vehicle audio | voice data, such as the siren sound which the emergency vehicle of the said local government blows for every local government Audio data 62 is included. And the emergency vehicle audio | voice data 62 has memorize | stored the audio data of the emergency vehicle in the form of the sonagram similarly to FIG.

  10 includes a car navigation system 80. The specific speech recognition apparatus 100-1 shown in FIG.

  The car navigation system 80 includes a GPS receiver 40-1 and a CPU 82 that performs arithmetic processing of the car navigation system 80, and includes a map data storage unit 50-1 connected to the CPU 82.

  The GPS receiver 40-1 includes a GPS antenna 42-1 that receives radio waves from GPS satellites. Moreover, the GPS receiver 44-1 which calculates the present position data of the specific speech recognition apparatus 100-1 as, for example, latitude and longitude data based on the radio wave received from the GPS antenna 42-1 is included. The GPS receiver 44-1 sends the calculated current position data to the CPU 82.

  The CPU 82 is connected to a ROM 84 that stores a program for operating the CPU 82, emergency vehicle approach warning image data, emergency vehicle approach warning message data, and the like, and a RAM 86 that stores various data as a work area of the CPU 82. The program stored in the ROM 84 and operating the CPU 82 includes a program for executing the function of the car navigation and the function of the current position specifying unit 32 described in the second embodiment.

  The CPU 82 is also connected to a display 72-1 for performing car navigation and a speaker 74-1 for outputting sound.

  The map data storage unit 50-1 includes, for example, map data 52-1 for storing map data for the whole of Japan together with latitude and longitude data. The map data 52-1 further stores information on which local government in Japan (□□ prefecture, XX city, △ Δ town, etc.) the latitude and longitude points belong to.

  The CPU 37-1 described above and the CPU 82 of the car navigation system 80 are connected by a bus (not shown) so that various types of information can be transmitted and received mutually.

  Next, the operation of the fourth embodiment will be described with reference to FIG.

  In the fourth embodiment of the present invention, a part of the components of the third embodiment shown in FIG. 4 is replaced with a car navigation system. Therefore, the operation shown in FIG. 11 is almost the same as the operation shown in FIG. 9. In FIG. 11, the operation shown in FIG. 9 is given the same reference numeral or symbol, and the description thereof is omitted as much as possible. .

  FIG. 11 is a sequence diagram for explaining the operation of the third embodiment.

  In FIG. 11, the microphone 10 of the specific speech recognition apparatus 100-1 acquires the sound around the microphone 10 in a unit of a certain time length and sends it to the sound data extraction unit 20 (step S <b> 1 in FIG. 11). After that, step S1 is repeated to acquire the next time length voice.

  The audio data extraction unit 20 amplifies the analog audio data sent from the microphone 10 with the amplifier 22 and sends it to the bandpass filter 24. The band-pass filter 24 passes the audio data only in the frequency range (for example, 250 to 900 Hz) of the siren sound that the emergency vehicle blows and sends it to the A / D converter 26. The A / D converter 26 converts the analog audio data received from the bandpass filter 24 into digital audio data (step S2-1). Then, the converted digital audio data is sent to the control unit 30.

  On the other hand, the GPS receiving unit 40-1 of the car navigation system 80 of the specific speech recognition apparatus 100-1 receives the radio waves transmitted from the plurality of GPS satellites by the GPS antenna 42-1 at regular time intervals. Then, based on the radio wave information received by the GPS antenna 42-1, the GPS receiver 44-1 calculates the current position of the specific speech recognition device 100-1 (step S3-2). Then, the calculated current position information is sent to the CPU 82. Thereafter, step S3-2 is repeated to receive radio waves from the GPS satellite at the next time interval.

  The CPU 82 of the car navigation system 80 that has received the current position from the GPS receiver 44-1 of the GPS receiver 40-1 causes the program of the current position specifying unit (current position specifying unit 32 in FIG. 2) to be executed. Then, the CPU 82 specifies which municipality the received current position corresponds to by referring to the map data 52-1 of the map data storage unit 50-1 shown in FIG. Further, the CPU 82 specifies other 2-3 local governments in the vicinity of the specified local government (step S4-2). And CPU82 sends out the information of the specified some local government to CPU37-1 of the control part 30. FIG.

  The CPU 37-1 that has received the specified plurality of local government information from the CPU 82 causes the emergency vehicle voice data selection unit (emergency vehicle voice data selection unit 34 in FIG. 2) to execute a program. And the emergency vehicle audio | voice data which the some local government specified by step S4-2 uses are extracted from the emergency vehicle audio | voice data 62 of the emergency vehicle audio | voice data storage part 60 shown in FIG. 10 (step S5-2). The number of emergency vehicle voice data extracted here is the number of types of emergency vehicles in a plurality of local governments, and the emergency vehicle voice data is three-dimensional data in the form of a sonogram.

  When the CPU 37-1 of the control unit 30 receives the digital audio data from the A / D converter 26 of the audio data extraction unit 20, the CPU 37-1 causes the alarm sound recognition unit (the alarm sound recognition unit 36 in FIG. 2) to execute a program. The CPU 37-1 first converts the digital audio data into three-dimensional data in the form of sonagram (referred to as received sonargram data) and stores it in the RAM 39-1. Next, the CPU 37-1 compares whether the received sonogram data matches any of the plurality of emergency vehicle voice data of the plurality of local governments extracted in step S5-2 (step S6-2). Then, the CPU 37-1 determines whether the received sonogram data matches any of the plurality of emergency vehicle voice data of the plurality of local governments extracted in step S5-2 (step S7-2). The coincidence is determined by pattern matching between the data represented by the sonagram. The CPU 37-1 returns the determination result (“match” or “non-match”) to the CPU 82 of the car navigation system 80.

  The CPU 82 of the car navigation system 80 determines whether the determination result is “match” or “mismatch” (step S71-2).

  When the determination result in step S71-2 in FIG. 11 is “match” (“match” in step S71-2), the CPU 82 of the car navigation system 80 determines that the emergency vehicle is approaching. Then, emergency vehicle approach warning image data, emergency vehicle approach warning message data, and the like indicating that an emergency vehicle is approaching are read from the ROM 84 and notified to the display 72-1 and the speaker 74-1. The display 72-1 outputs and notifies that the emergency vehicle is approaching as image data. At the same time or / or the speaker 74-1 outputs and notifies that the emergency vehicle is approaching (step S8-2). When the user of the specific speech recognition apparatus 100-1 notices this notification and performs a notification stop operation, the display 72-1 and the speaker 74-1 stop the notification and return to the original state.

  When the determination result in step S71-2 is “mismatch” (step S71-2 “mismatch”), the CPU 82 of the car navigation system 80 does nothing and returns to the original state.

  The operation of the fourth embodiment of the present invention has been described above.

  As described above, the fourth embodiment of the present invention is configured so that the car navigation system 80 replaces some of the components of the third embodiment shown in FIG.

  Therefore, according to the present embodiment, the following effects similar to those in the third embodiment can be obtained.

  That is, even if the emergency vehicle audio data (siren sound, etc.) used by the local government is different for each local government, the audio data outside the vehicle matches the emergency vehicle audio data generated by the emergency vehicle. Can be determined. And when it corresponds, it becomes possible to alert | report in the vehicle that the emergency vehicle is approaching.

  Further, the band-pass filter 24 can extract only voice data in a frequency range where the emergency vehicle blows, and the received voice data can be efficiently converted into a sonogram.

  Further, the emergency vehicle audio data 62 stores a large number of audio data that sounds from emergency vehicles of all local governments. Only vehicle audio data can be extracted. Therefore, it is possible to greatly reduce the number of pattern matching of the data form of the sonogram.

  In addition, according to the fourth embodiment, since the existing car navigation system 80 is used, the cost of the specific speech recognition apparatus 100-1 can be reduced.

In each of the above embodiments, the siren sound that the emergency vehicle blows mainly has been described as an example of the region-specific sound, but the present invention is not limited to this. Each of the above embodiments can be applied to speech unique to a geographical region. Moreover, although the local government was demonstrated as a representative example as a geographical area, it is not restricted to this. If the area-specific voice is heard in advance, it can be regarded as a geographical area and each embodiment can be applied.

2 Voice acquisition unit 3-2 Current position recognition specifying unit 3-6 Voice recognition result notification unit 4 Current position generation unit 6 Area specific voice storage unit 10 Microphone 20 Audio data extraction unit 22 Amplifier 24 Band pass filter 26 A / D converter 30 Control unit 32 Current position specifying unit 34 Emergency vehicle voice data selection unit 36 Alarm sound recognition unit 37 CPU
37-1 CPU
38 ROM
38-1 ROM
39 RAM
39-1 RAM
40 GPS receiver 40-1 GPS receiver 42 GPS antenna 42-1 GPS antenna 44 GPS receiver 44-1 GPS receiver 50 Map data storage unit 50-1 Map data storage unit 52 Map data 52-1 Map data 60 Emergency vehicle sound Data storage unit 62 Emergency vehicle voice data 70 Alarm output unit 72 Display 72-1 Display 74 Speaker 74-1 Speaker 80 Car navigation system 82 CPU
84 ROM
86 RAM
100 specific speech recognition device 100-1 specific speech recognition device 600 sonagram pattern data (first example)
700 sonagram pattern data (second example)

Claims (8)

An audio acquisition unit that acquires input peripheral audio in units of a certain time length and outputs it as digital audio data;
Based on a predetermined signal that is input, a current position generation unit that calculates and outputs its current position;
A region-specific speech storage unit that stores region-specific speech heard only in a predetermined geographical region in association with the geographical region in units of the predetermined time length or more;
A current position recognition specifying unit that outputs information indicating a specific geographical area to which the current position belongs, which is input from the current position generating unit;
If the region specific voice extracted from the region specific voice storage unit for the specific geographical region to be input is compared with the digital voice data input from the voice acquisition unit, and it is determined that both match, A specific speech recognition apparatus comprising: a speech recognition result notifying unit that outputs a notification signal for notifying the user. The voice acquisition unit is configured to acquire a peripheral voice in a unit of a predetermined time length;
An audio data extraction unit that A / D-converts analog audio data acquired by the microphone and sends it to the control unit as digital audio data;
The current position generation unit includes a GPS reception unit that receives radio waves transmitted by a plurality of GPS satellites at regular time intervals, calculates a current position based on the received radio wave information, and transmits the current position to the control unit,
The area-specific voice storage unit is connected to the control unit, and includes, for each municipality, an emergency vehicle voice data storage unit that stores emergency vehicle voice data such as a siren sound that the emergency vehicle of the municipality blows.
The current position recognition specifying unit is connected to the control unit, and includes a map data storage unit for storing map data, and a plurality of near the current position based on the current position received from the GPS receiving unit. A current position specifying unit that specifies the local government with reference to the map data storage unit,
The voice recognition result notifying unit is included in the control unit, and emergency vehicle voice data for extracting emergency vehicle voice data used by a plurality of local governments specified by the current position specifying unit from the emergency vehicle voice data storage unit A selection section;
The digital voice data included in the control unit and received from the voice data extraction unit is determined to match any of a plurality of emergency vehicle voice data of a plurality of local governments extracted by the emergency vehicle voice data selection unit, An alarm sound recognizing unit for notifying an alarm output unit connected to the control unit that an emergency vehicle is approaching,
The warning output unit, when notified from the warning sound recognition unit, outputs and informs that an emergency vehicle is approaching by voice, message data, or image data,
The specific speech recognition apparatus according to claim 1. The voice data extraction unit includes a band-pass filter, and the band-pass filter extracts only voice data in a frequency range of emergency vehicle voice data such as a siren sound generated by a local emergency vehicle.
The specific speech recognition apparatus according to claim 2. The audio data stored in the emergency vehicle audio data storage unit is audio data in the form of a sonogram,
The warning sound recognizing unit converts the digital sound data received from the sound data extracting unit into sound data in the form of a sonogram, and then a plurality of emergency vehicle sound data of a plurality of local governments extracted by the emergency vehicle sound data selecting unit. It is determined by pattern matching whether it matches any of
The specific speech recognition apparatus according to claim 2, wherein the specific speech recognition apparatus is provided. The map data is mounted on a car navigation system, and the car navigation system executes the functions of the GPS receiver, the current position specifying unit, and the alarm output unit.
The specific speech recognition apparatus according to any one of claims 2 to 4, wherein the specific speech recognition apparatus is characterized. Acquire surrounding audio as analog audio data in units of a certain length of time,
Converting the analog audio data into digital audio data by A / D conversion;
Receive radio waves sent by multiple GPS satellites at regular time intervals, calculate the current position based on the received radio wave information,
Based on the current location, identify a plurality of local governments near the current location,
Extracting a plurality of emergency vehicle voice data used by the plurality of specified local governments,
It is determined whether the digital voice data matches any of a plurality of emergency vehicle voice data of a plurality of local governments in the vicinity of the current position, and if determined to match, voice, message data indicating that an emergency vehicle is approaching, Alternatively, output and notify by image data,
A specific speech recognition method characterized by the above. The A / D conversion is performed after extracting only the voice data in the frequency range of the emergency vehicle voice data such as sirens sounded by the emergency vehicle of the local government from the analog voice data.
The specific speech recognition method according to claim 6. The emergency vehicle sound data is sound data in the form of a sonagram;
After converting the digital voice data into voice data in the form of a sonogram, it is determined by pattern matching whether it matches any of a plurality of emergency vehicle voice data of a plurality of local governments in the vicinity of the current position.
The specific speech recognition method according to claim 6, wherein the specific speech recognition method is provided.

Images