JP2008298881A - Speech reproducing apparatus and speech reproducing method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To suppress reproduction of harsh (uncomfortable) speech at the time of error, by reducing a noise floor of 16kbps adaptive differential pulse code modulation (ADPCM) and noise at the time of wireless error to a 32kbps ADPCM level. <P>SOLUTION: A speech reproducing apparatus comprises: a data receiving section 250 for receiving a coded speech data; a data determination section 254 for determining whether the coded speech data is a low speed data of 16kbps, or a high speed data which is faster than 16kbps; a first decoder 256 which decodes the coded speech data to a PCM data, when it is the high speed data; a second decoder which loads and decodes the coded speech data to the PCM data, when it is the low speed data; a digital filter 274 which suppresses a high frequency region of the decoded PCM data by the second decoder; a speech conversion section 258 which converts the PCM data to an analog speech signal; and a speech reproducing section 260 which reproduces the analog speech signal. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to an audio reproducing apparatus and an audio reproducing method for reproducing encoded audio data encoded using an audio codec.

  In a wireless terminal such as PHS (Personal Handy phone System), an analog signal based on a user's voice is once converted into a digital signal and transmitted to another user's wireless terminal, where it is converted again into an analog signal to reproduce the voice. . However, simply converting an analog signal into a digital signal increases the amount of information, and cannot transmit all audio data through a communication path in which the amount of information that can be transmitted is limited. Therefore, an audio codec that compresses the amount of information of audio data is necessary so that it can be transmitted through this communication path.

  The audio codec is composed of encoding (decoding) and decoding (decoding) of audio digital data, and audio data encoded by the audio encoding device is decoded and reproduced by the audio reproducing device. Examples of the voice codec used in the above-described wireless terminal such as PHS include ADPCM (Adaptive Differential PCM) coding technology. Such ADPCM is a technique that encodes a difference of audio data by utilizing the property that analog data such as audio continuously changes.

  In a wireless terminal, for example, 32 kbps ADPCM (ITU-T (International Telecommunication Union-Telecommunication sector) recommendation G.726) is adopted in a QPSK (Quadrature Phase Shift Keying) modulation method, and here, reproduction without a sense of incongruity in hearing is adopted. Therefore, it is possible to generate audio data that maintains continuity between the current frame data arranged on the time axis and the previous frame data by performing windowing using a window function.

  In the wireless terminal, the parallel use of 16 kbps ADPCM in the BPSK modulation scheme is being studied in addition to the above-described 32 kbps ADPCM in the QPSK modulation scheme from the viewpoint of effective use of the radio channel. Although 16 kbps ADPCM transmits less information per unit time than 32 kbps ADPCM, it has an advantage that communication can be performed relatively stably even when the radio wave intensity is weak.

Here, the operation by ADPCM will be briefly described. As shown in FIG. 9, the speech encoding apparatus 10 determines the difference d _n between the PCM data PCM to be transmitted and the previous PCM data predicted value PCM ′ _n−1. A code is assigned after being quantized with a quantization width Δ _n , and the encoded encoded audio data (ADPCM data) ADPCM _n is transmitted to the audio reproduction device 12. In the speech encoding apparatus 10, the encoded speech data ADPCM _n is further subjected to inverse adaptive quantization, and the difference d ′ _n subjected to inverse adaptive quantization is added to the previous value PCM ′ _n−1 to prepare for the next time. Then, the audio reproduction device 12 performs inverse adaptive quantization on the encoded audio data ADPCM _n using the quantization width Δ _n , and adds the inverse adaptive quantized difference d ′ _n to the previous value PCM ′ _n−1 . PCM data PCM is obtained.

Thus, ADPCM has a characteristic that the resolution of quantization can be changed in accordance with the amplitude change width so that even analog data having a small amplitude can be reproduced, and in accordance with encoded speech data ADPCM _n as shown in FIG. Thus, the quantization width Δ _n is increased where the amplitude range of the analog data is large, and the quantization width Δ _n is decreased where the amplitude range is small. By using such ADPCM, the PCM data PCM _n of the speech coding apparatus 10 can be substantially restored from a small amount of compressed information.

  However, when such an ADPCM decoding circuit is applied to 16 kbps ADPCM, an abnormal noise sound may occur when an error occurs due to its characteristics, and an unpleasant sound may be heard. This is because when a continuous error occurs in the encoded speech data, the original PCM value is far from the calculated value, and it takes time until the original value is restored in the 16 kbps ADPCM with a small amount of difference data.

  Therefore, in the decoding circuit, there is a technique for further encoding the PCM data obtained by decoding the encoded audio data and initializing the decoding circuit when the difference between the encoded signal and the input signal exceeds a predetermined threshold. It is disclosed (for example, Patent Document 1).

However, since the above-described technique requires two decoding circuits to compare the encoded signal and the input signal, the cost is increased, and the signal and input signal encoded by the initialization condition of the decoding circuit are increased. Since it is determined only from the difference between the two, high-precision initialization could not be expected.
JP-A-1-179522

  As described above, the 16 kbps ADPCM is inferior to the 32 kbps ADPCM in the following two points due to the small amount of information of transmission data when no processing is performed.

  The first is the transition of the noise floor. For example, FIG. 10 shows a noise floor when a 1 kHz sine wave is input at a common maximum reception volume in 16 kbps ADPCM and 32 kbps ADPCM. The 16 kbps ADPCM has a larger quantization noise than the 32 kbps ADPCM, and when the received sound volume at 1 kHz is adjusted to -20 dB, the noise floor will increase by about 13.2 dB on average.

  The second problem is noise caused by a radio error. For example, FIG. 11 shows a sound pressure distribution when a single tone (pin tone) having a constant volume of 1600 Hz and −16 dB is transmitted to 16 kbps ADPCM and 32 kbps ADPCM and received at a FER (Frame Error Rate) equivalent to the movement test. ing. In FIG. 11, the horizontal axis represents the sound pressure (dB), and the vertical axis represents the number of samples of the sound pressure. Therefore, although the number of samples of −16 dB, which is the same as that of a single sound, is maximized, other sound pressures are also sampled due to an error or a handover for avoiding interference.

  Referring to FIG. 11, it can be understood that the sound pressure distribution of 16 kbps ADPCM is higher than that of 32 kbps ADPCM particularly at high sound pressure. This indicates that the 16 kbps ADPCM has a very high value for both noise level and frequency compared to the 32 kbps ADPCM. When the sound pressure of the noise increases, a violent sound “buzz” is mixed into the voice, which makes the user uncomfortable.

  The present invention has been made in view of the above problems at the time of audio data reproduction in 16 kbps ADPCM, and an object of the present invention is to reduce the noise floor of 16 kbps ADPCM and the noise at the time of a radio error to 32 kbps ADPCM level, thereby causing an annoyance at the time of error ( It is an object to provide a new and improved audio reproduction device and audio reproduction method capable of suppressing unpleasant sound reproduction.

  In order to solve the above-described problem, according to one aspect of the present invention, a data receiving unit that receives encoded audio data encoded using an audio codec based on ADPCM (adaptive differential PCM), and encoded audio data include: A data determination unit that determines whether low-speed data of 16 kbps or high-speed data other than 16 kbps and higher than 16 kbps; a first decoder that decodes encoded audio data into PCM data when the encoded audio data is high-speed data; A second decoder that loads the encoded voice data and decodes it into PCM data when the encoded voice data is low-speed data; a digital filter that suppresses at least a high frequency of the decoded PCM data; and a first decoder Alternatively, an audio conversion unit that converts PCM data from the digital filter into an analog audio signal, and an analog Characterized in that it comprises a sound reproduction unit for reproducing audio signal, and audio reproduction apparatus is provided.

  For 16 kbps ADPCM, if decoding is performed in the same way as 16 kbps ADPCM, for example, at a speed other than 16 kbps, such as 32 kbps ADPCM, the problem of noise floor occurs as described above due to the small amount of information of differentially encoded audio data. In the present invention, the encoded audio data is temporarily converted into software and decoded. At that time, the noise floor can be reduced by suppressing the high band with a digital filter, and a noise floor equivalent to 32 kbps ADPCM can be achieved.

  The audio reproduction device is provided in the preceding stage of the digital filter and converts the PCM data into linear PCM data with an extended bit length, and the linear PCM data that is provided in the subsequent stage of the digital filter and passes through the digital filter. And a PCM conversion unit that converts the data into PCM data.

  The input of the digital filter described above is preferably linear PCM data in which the scale of the PCM data is fixed. With the configuration of the linear conversion unit and the PCM conversion unit, desired input data for the digital filter can be generated, and resources such as an existing audio conversion unit can be used as they are by returning the data format to the original PCM data. it can.

  The audio reproduction device includes: a noise determination unit that determines whether the PCM data from the second decoder is noise; and a reset unit that resets a previous value held by the second decoder when the noise determination unit determines noise. , May be further provided.

  If errors continue, the previous value held by the second decoder diverges in one direction, and the actual audio data and the estimated audio data are separated. Therefore, since the 16 bps ADPCM decoder takes in the difference from the previous value of the PCM data, it takes time to recover the original normal data once the PCM data becomes an incorrect value. Therefore, when the noise determination unit detects noise, it abandons maintenance of the previous value of the second decoder and restarts decoding. In this way, it is possible to prevent the previous value from diverging due to noise, and it is possible to prevent the output of abnormal noise sounds and unpleasant sounds due to the occurrence of errors.

  The sound reproduction device further includes a cutoff frequency changing unit that changes the cutoff frequency according to the noise determination of the noise determination unit, and sets the cutoff frequency lower than when it is not determined as noise while it is determined as noise. May be.

  With this configuration, it is possible to suppress a high band when noise is generated, and to further reduce the noise floor. For example, when the cutoff frequency when it is not determined as noise is 3000 Hz, the time when it is determined as noise is reduced to 2000 Hz. The frequency at which a person is likely to feel sound pressure is in the vicinity of 4000 Hz, and by suppressing signals of 2000 Hz or higher, it is possible to further suppress harsh voices.

  If the noise determination unit determines that the noise is noise, an attenuation unit that attenuates the PCM data to 1/8 may be further provided.

  When the noise determination unit determines that the PCM data is noise, it is not necessary to input the noise to the digital filter. However, if the PCM data is completely silent, a silent state is intermittently generated during reproduction, which causes a sense of incongruity. In the present invention, about 1/8 of the PCM data is left unintentionally to maintain good hearing.

  In order to solve the above problems, according to another aspect of the present invention, a data receiving step for receiving encoded audio data encoded using an audio codec based on ADPCM, and low-speed data with encoded audio data of 16 kbps Or a data determination step that determines whether the data is faster than 16 kbps and other than 16 kbps, and if the determination in the data determination step is high-speed data, a first decoding step that decodes encoded audio data into PCM data, A second decoding step for loading the encoded voice data and decoding the PCM data, a digital filter step for suppressing at least a high frequency of the PCM data decoded in the second decoding step, and converting the PCM data into an analog voice signal. Audio conversion step to convert and playback of analog audio signal Characterized in that it includes a sound reproducing step, and audio reproduction method is provided.

  The components corresponding to the technical idea of the above-described sound reproduction device and the description thereof can be applied to the sound reproduction method.

  As described above, according to the present invention, it is possible to reduce the noise floor of 16 kbps ADPCM and the noise at the time of radio error to 32 kbps ADPCM level, and to suppress the reproduction of unpleasant (unpleasant) sound at the time of error. Therefore, compared with 32 kbps ADPCM, it is possible to reduce a wide call area and error rate, and to maintain a sound quality level equivalent to 32 kbps ADPCM.

  Exemplary embodiments of the present invention will be described below in detail with reference to the accompanying drawings. In addition, in this specification and drawing, about the component which has the substantially same function structure, duplication description is abbreviate | omitted by attaching | subjecting the same code | symbol.

  For example, 32 kbps ADPCM is used as an audio codec for digital data in communication between a wireless terminal such as a PHS terminal or a mobile phone and a base station. Further, in recent years, parallel use of 16 kbps ADPCM in addition to this 32 kbps ADPCM has been studied from the viewpoint of effective use of a wireless line. Due to the characteristics of the 16 kbps ADPCM, noise is generated when an error occurs, and an unpleasant voice may be heard.

(Wireless communication system 100)
In the present embodiment, there is provided an audio reproduction device capable of reliably preventing reproduction of unpleasant (unpleasant) audio at the time of an error even in such a 16 kbps ADPCM. In the present embodiment, the wireless terminal 110 is used as an audio playback device. Hereinafter, an outline of the wireless communication system 100 that performs wireless communication by the wireless terminal 110 as the sound reproducing device in the present embodiment will be described, and then each component will be described in detail.

  FIG. 1 is a block diagram illustrating a schematic configuration of a wireless communication system 100. The wireless communication system 100 includes a wireless terminal 110, a base station 120, a communication network 130 such as an ISDN (Integrated Services Digital Network) line, the Internet, a dedicated line, a relay server 140, and another wireless terminal 150. Consists of.

  In the wireless communication system 100, when attempting to make a call to another wireless terminal 150 using the wireless terminal 110, the user operates his / her wireless terminal 110 to communicate with the base station 120 in the wireless communication available area. Establish wireless communication, and other communication partners have through the communication network 130, the relay server (including the base station control device and the exchange) 140, and the base station 120 in the wireless communication area of the other wireless terminal 150. A voice call is performed with the wireless terminal 150.

(Wireless terminal 110)
FIG. 2 is a functional block diagram illustrating a hardware configuration of the wireless terminal 110. The wireless terminal 110 includes various electronic devices such as a notebook personal computer and a PDA (Personal Digital Assistant) in addition to the above-described mobile phone and PHS. As components, a terminal control unit 210, a terminal memory 212, , A display unit 214, an operation unit 216, a voice input / output unit 218, and a terminal radio unit 220.

  The terminal control unit 210 manages and controls the entire wireless terminal 110 using a semiconductor integrated circuit including a central processing unit (CPU). The terminal control unit 210 naturally performs a call function and a mail distribution function using the wireless terminal 110 using the program in the terminal memory 212, but a second decoder 262, linear conversion, which will be described later with reference to FIG. It also functions as a unit 264, a noise determination unit 266, a cutoff frequency changing unit 268, an attenuation unit 270, a reset unit 272, a digital filter 274, and a PCM conversion unit 276.

The terminal memory 212 includes a ROM, a RAM, an E ² PROM, a nonvolatile RAM, a flash memory, a HDD (Hard Disk Drive), and the like, and stores a program processed by the terminal control unit 210, audio data, and the like.

  The display unit 214 is configured by a color or monochrome display, and is stored in the terminal memory 212 or provided from an application relay server (not shown) via the communication network 130, or a GUI (Web GUI or application GUI). Graphical User Interface) can be displayed.

  The operation unit 216 includes switches such as a keyboard, a cross key, and a joystick, and receives a user operation input.

  The voice input / output unit 218 includes a microphone and a speaker, and converts a user's voice input during a call into voice data or changes voice data of a call partner into voice and outputs the voice data. Further, a ring tone, an operation sound by the operation unit 216, an alarm sound, and the like can be output. It also functions as a first decoder 256, an audio conversion unit 258, and an audio reproduction unit 260, which will be described later.

  The terminal radio unit 220 performs radio communication with the base station 120 in the mobile phone network. Further, it also functions as a data receiving unit 250, a receiving buffer 252, and a data determining unit 254 described later.

  The hardware of the wireless terminal 110 has been described above with reference to FIG. 2. The functions and operations performed on such hardware will be described in detail below.

  FIG. 3 is a functional block diagram illustrating schematic functions of the wireless terminal 110. The wireless terminal 110 includes a data reception unit 250, a reception buffer 252, a data determination unit 254, a first decoder 256, an audio conversion unit 258, an audio reproduction unit 260, a second decoder 262, and a linear conversion unit 264. A noise determination unit 266, a cut-off frequency changing unit 268, an attenuation unit 270, a reset unit 272, a digital filter 274, and a PCM conversion unit 276.

  The data receiving unit 250 receives encoded speech data (hereinafter referred to as ADPCM data) encoded using an ADPCM (adaptive differential PCM) speech codec, specifically 16 kbps ADPCM and 32 kbps ADPCM, every 5 msec. Store in the buffer 252. For example, in the case of 16 bps ADPCM, data of 10 bytes (80 bits = 40 symbols) is received every 5 msec. In this embodiment, 16 kbps is described as low-speed data and 32 kbps is described as high-speed data. However, high-speed data is not limited to this, and other than 16 kbps, such as 24, 32, 48, 64,. Includes a variety of faster ADPCMs.

  The reception buffer 252 temporarily holds ADPCM data received by the data receiving unit 250.

  The data determination unit 254 determines whether the ADPCM data held in the reception buffer 252 conforms to the 32 kbps ADPCM codec or the 16 kbps codec. If the determined result is 32 kbps ADPCM, the ADPCM data is transmitted to the existing first decoder 256, and if it is 16 kbps ADPCM, the ADPCM data is transmitted to the second decoder 262 according to the present embodiment.

  The first decoder 256 acquires 32 kbps ADPCM data from the reception buffer 252 through the data determination unit 254, and decodes the ADPCM data into PCM data. Such PCM data includes audio of 300 to 3300 Hz. In the present embodiment, an existing integrated circuit decoder is assumed as the first decoder 256, but it may be realized by a programmable element such as an FPGA (Field Programmable Gate Array) or an ASIC (Application Specific Integrated Circuit) or software. Is possible.

  The audio conversion unit 258 temporarily converts the PCM data decoded by the first decoder into linear PCM data, and further generates an analog audio signal through a D / A converter.

  The audio reproduction unit 260 reproduces the analog audio signal from the audio conversion unit 258. The data reception unit 250, the reception buffer 252, the first decoder 256, and the audio conversion unit 258 described above are already provided in the wireless terminal 110 for reproducing 32 kbps ADPCM audio, and are necessary for reproducing the 32 kbps ADPCM audio data. Use existing components. Therefore, the same high level sound quality as in the related art can be maintained for 32k ADPCM.

  The second decoder 262 is a decoder corresponding to the 32 kbps ADPCM first decoder 256, and is different from the first decoder 256 in that digital processing on software is performed. The second decoder 262 acquires 16 kbps ADPCM data from the reception buffer 252 through the data determination unit 254, and decodes the ADPCM data into PCM data. At this time, the frequency range of the PCM data is 300 to 3300 Hz.

  The linear conversion unit 264 converts PCM data into linear PCM data with an extended bit length for a digital filter described later. For this conversion, μ-law or A-law conversion for converting 8-bit PCM data into 16-bit linear PCM data is used. The μ-law or A-law conversion is one of the compression / decompression methods in speech coding standardized by ITU-T. In addition, μ-law is used in Japanese ISDN networks.

  The input of the digital filter 274, which will be described later, is preferably linear PCM data in which the scale of the PCM data is fixed. With the configuration of the linear conversion unit 264 and the PCM conversion unit 276, which will be described later, desired input data to the digital filter 274 can be generated, and resources such as the existing audio conversion unit 258 are restored by returning the data format to the original PCM data Can be used as is.

The noise determination unit 266 determines whether the linear PCM data converted by the linear conversion unit 264 is noise. As errors detected in the present embodiment, confirmation of whether or not the CI information indicating the type of frame indicates voice data (TCH), CRC (Cyclic Redundancy Check), presence or absence of a synchronization code (UW: Unique Word), The confirmation of whether or not the upper limit (for example, 2 ¹⁶ ) of the linear PCM data is exceeded is used. In addition to the errors listed here, various error detection methods that can determine the validity of the frame can be applied.

  The cut-off frequency changing unit 268 changes the cut-off frequency of a digital filter 274, which will be described later, according to the presence or absence of noise from the noise determination unit 266. Therefore, when the noise determination unit 266 determines that the noise is present, the cut-off frequency is set lower than when it is not determined. For example, when the cutoff frequency is 3000 Hz in a normal state when there is no noise, the cutoff frequency is lowered to about 2000 Hz when there is noise.

  When the noise determination unit 266 determines that the noise is noise, the attenuation unit 270 attenuates the linear PCM data to 1/8 (−18 dB). When the noise determination unit 266 determines that the PCM data is noise, it is not necessary to input the noise to the digital filter 274. However, if the PCM data is completely silent, a silent state is intermittently generated during reproduction, which causes a sense of incongruity. In the present embodiment, about 1/8 of the PCM data is left dare to maintain good audibility.

  When the noise determination unit 266 determines that the noise is noise, the reset unit 272 resets the previous value held by the second decoder 262.

  If errors continue, the previous value held by the second decoder 262 diverges in one direction, and the actual audio data and the estimated audio data are separated. Therefore, since the 16 bps ADPCM decoder takes in the difference from the previous value of the PCM data, it takes time to recover the original normal data once the PCM data becomes an incorrect value. Therefore, when the noise determination unit 266 detects noise, the second decoder 262 abandons maintenance of the previous value and newly restarts decoding. In this way, it is possible to prevent the previous value from diverging due to noise, and it is possible to prevent the output of abnormal noise sounds and unpleasant sounds due to the occurrence of errors.

  The digital filter 274 suppresses at least a high frequency of the decoded linear PCM data and transmits the suppressed linear PCM data to the audio conversion unit 258. The frequency band of linear PCM data input to the digital filter 274 is 300 to 3300 Hz. By suppressing the high frequency signal of the linear PCM data, for example, a signal of 3000 Hz or higher, audio data in an unnecessary region of 3000 Hz or higher can be reduced. Therefore, the high band of the noise floor can be set to the same level as 32 kbps, and the audibility can be improved. Such filtering is performed at a cycle of 8000 Hz and oversampling 8 times.

  In order to realize the cutoff frequency as described above, a bandpass filter BPF of 300 to 3000 Hz is used in the present embodiment. Moreover, when the noise determination part 266 detects noise, a pass band is narrowed to 300-2000 Hz, and the high frequency signal is further suppressed. Here, the cutoff frequency of the high-frequency cut is set to 3000 Hz when there is no noise, and 2000 Hz when there is no noise. However, it is not always necessary to set this value, and it is 3000 to 3400 Hz when there is no noise, and 1800 to 2200 Hz if there is Can be set arbitrarily.

  With this configuration, it is possible to suppress a high band when noise is generated, and to further reduce the noise floor. The frequency at which a person is likely to feel sound pressure is in the vicinity of 4000 Hz, and by suppressing signals of 2000 Hz or higher, it is possible to further suppress harsh voices.

  Further, such a band pass filter may be constituted by a Kaiser window among the window functions. The Kaiser window can be adjusted for bandpass shape and window length by changing the coefficients in its window function. As described above, by forming a Kaiser window by software as the digital filter 274 of the present embodiment, a band pass filter having remarkable characteristics that cannot be achieved by hardware can be configured. In addition, since the software does not change parameters due to temperature or time, it can generate stable and high-quality sound. When such a filter is configured by hardware, the shape becomes large by sampling at 8000 Hz, and this leads to an increase in cost, which is not realistic, but if there is no problem in processing speed, a programmable element such as FPGA or ASIC can be used. It can also be realized.

  Such reset of the internal parameters of the digital filter 274 is performed in a silent state when switching to a new wireless communication by handover in the wireless terminal.

  The PCM conversion unit 276 is provided after the digital filter, converts linear PCM data that has passed through the digital filter into PCM data, and transmits the PCM data to the audio conversion unit 258 as in the case of 32 kbps ADPCM.

  With the configuration as described above, the wireless terminal 110 can reproduce ADPCM data encoded using a voice codec while avoiding annoying voice at the time of an error.

(Audio playback method)
Next, an audio reproduction method for reproducing ADPCM data using the wireless terminal 110 as the above-described audio reproduction apparatus will be described.

  FIG. 4 is a flowchart showing a specific flow of the audio reproducing method. In this audio reproduction method, first, ADPCM data encoded using an ADPCM audio codec is received (S300: data reception step), and it is determined whether the ADPCM data is 32 kbps or 16 kbps (S302). : Data judgment step).

  In the data determination step (S302), when it is determined to be 32 kbps, the ADPCM data is transmitted to the first decoder 256, and is decoded into the PCM data by the first decoder 256 (S304: first decoding step).

  If it is determined in the data determination step (S302) that the data rate is 16 kbps, the ADPCM data is loaded into the second decoder 262, decoded into PCM data by the second decoder 262 (S306: second decoding step), and the decoded PCM data. Is converted into 16-bit linear PCM data by μ-law conversion (S308: linear conversion step).

  Here, noise determination is performed on the linear PCM data (S310: noise determination step), and it is confirmed that the linear PCM data is regular audio data and is not noise. Specific processing of the noise determination step (S310) will be described in detail later. When it is not determined as noise in the noise determination step (S310), the passband of the bandpass filter is set to 300 to 3000 Hz (S312), and digital filtering by the Kaiser window is executed (S314: digital filter step).

  When it is determined as noise in the noise determination step (S310), the passband of the bandpass filter is set to 300 to 2000 Hz (S320), the amplitude of the linear PCM data is attenuated to 1/8 (S322: attenuation step), The previous value held by the second decoder 262 is reset (S324: reset step). Then, digital filtering is performed using the modified Kaiser window with a passband of 300 to 2000 Hz (S326: digital filter step). Specific processing of the digital filtering steps (S314 and S326) will be described in detail later.

  Then, the linear PCM data is converted again into 8-bit PCM data in order to match the sound reproduction path of 32 kbps (S328: PCM conversion step).

  At this point, since 8-bit PCM data is created in either case of 16 kbps or 32 kbps, the voice conversion unit 258 converts the PCM data into an analog voice signal (S330: voice conversion step), and finally An analog audio signal is reproduced (S332: audio reproduction step).

(Noise determination step: S310)
FIG. 5 is a flowchart showing a detailed flow of the noise determination step in the audio reproduction method. The noise determination unit 266 determines whether the linear PCM data converted by the linear conversion unit 264 is noise. Here, after concatenating PCM data packets in units of frames, an error is detected for each frame. Specifically, it is determined whether or not the CI information indicating the frame type indicates voice data (TCH) (S350). If the frame is voice data, the presence or absence of a bit error is determined by CRC ( S352). Here, it can also be determined whether or not the frame format, sampling frequency, transmission rate, etc. show reasonable numerical values. If no bit error is detected even in the CRC, the presence / absence of a synchronization code UW, which is represented by a specific bit string and indicates the position of ADPCM data, is determined (S354). When the synchronization code UW is detected at a normal position, it is determined whether or not the linear PCM data exceeds the upper limit (for example, 2 ¹⁶ ) (S356).

  If all the determination criteria are satisfied in such a noise determination step, the linear PCM data is regarded as “not noise”, and if none is satisfied, it is determined as “noise”.

(Digital filter step: S314, S326)
FIG. 6 is a flowchart showing a detailed flow of the digital filter step in the audio reproduction method. First, the 16-bit linear PCM data Sample (t) at time t is multiplied by 8 and substituted into the filter delay Tap (t) (S370). Then, by multiplying each the Tap (t) to each of the coefficient K (i), further result by adding (S372), finally the result of addition to the return value is divided by ^{2 12} (S374). Here, each coefficient is 300 to 3000 Hz, and K (i) = {K (0),..., K (12)} = {5, −37, −13, 15, −93, 60, 343, 149, −95, −7, 5, −44, 0}, K (i) = {− 10, 10, −19, −67, −24, 122, 223, 159, 6, −68, − at 300 to 2000 Hz 33, 8, -2}, and is replaced whenever the frequency band changes. In such a digital filter step, a signal of 300 to 2000 Hz or 300 to 3000 Hz can be extracted by a band addition filter with a period of 8000 Hz and oversampling 8 times.

  Also in the audio reproduction method described above, it is possible to reduce the noise floor of 16 kbps ADPCM and the noise at the time of radio error to 32 kbps ADPCM level, and to suppress the reproduction of unpleasant (unpleasant) audio at the time of error.

  FIG. 7 shows a noise floor when a sine wave of 1 kHz is input at a common maximum reception volume in 16 kbps ADPCM and 32 kbps ADPCM when the present embodiment is applied, and FIG. The sound pressure distribution is shown when a single sound of a volume is transmitted and received by FER equivalent to the movement test. As can be seen with reference to this figure, when this embodiment is applied, it is possible to approach 32 kbps ADPCM compared to the conventional FIGS. 10 and 11, and the advantages of 16 kbps ADPCM, that is, a wide call area and error rate. Reduction can also be achieved.

  As mentioned above, although preferred embodiment of this invention was described referring an accompanying drawing, it cannot be overemphasized that this invention is not limited to the example which concerns. It will be apparent to those skilled in the art that various changes and modifications can be made within the scope of the claims, and these are naturally within the technical scope of the present invention. Understood.

  Note that the steps in the audio reproduction method of the present specification do not necessarily have to be processed in time series in the order described in the flowchart, but are performed in parallel or individually (for example, parallel processing or object-based processing). Processing).

  The present invention can be used for an audio reproducing apparatus and an audio reproducing method for reproducing encoded audio data encoded using an audio codec.

1 is a block diagram showing a schematic configuration of a wireless communication system. It is the functional block diagram which showed the hardware constitutions of the radio | wireless terminal. It is the functional block diagram which showed the schematic function of the radio | wireless terminal. It is the flowchart which showed the specific flow of the audio | voice reproduction | regeneration method. It is the flowchart which showed the detailed flow about the noise determination step in an audio | voice reproduction method. It is the flowchart which showed the detailed flow about the digital filter step. It is the figure which compared the noise floor. It is the figure which compared sound pressure distribution. It is a block diagram for demonstrating an audio codec. It is the figure which compared the noise floor. It is the figure which compared sound pressure distribution.

Explanation of symbols

DESCRIPTION OF SYMBOLS 110 ... Wireless terminal 250 ... Data reception part 252 ... Reception buffer 254 ... Data determination part 256 ... 1st decoder 258 ... Audio | voice conversion part 260 ... Audio | voice reproduction | regeneration part 262 ... 2nd decoder 264 ... Linear conversion part 266 ... Noise determination part 268 ... Cutoff frequency changing unit 270 ... Attenuating unit 272 ... Reset unit 274 ... Digital filter 276 ... PCM conversion unit

Claims (6)

A data receiving unit for receiving encoded audio data encoded using an audio codec by ADPCM;
A data determination unit for determining whether the encoded audio data is low-speed data of 16 kbps or high-speed data other than 16 kbps and higher than 16 kbps;
A first decoder for decoding the encoded audio data into PCM data when the encoded audio data is high-speed data;
A second decoder for loading the encoded audio data and decoding it into PCM data when the encoded audio data is low-speed data;
A digital filter for suppressing at least a high frequency of the PCM data decoded by the second decoder;
An audio conversion unit for converting PCM data from the first decoder or the digital filter into an analog audio signal;
An audio reproduction unit for reproducing the analog audio signal;
An audio reproducing apparatus comprising: A linear conversion unit that is provided before the digital filter and converts the PCM data into linear PCM data with an extended bit length;
A PCM conversion unit that is provided at a subsequent stage of the digital filter and converts linear PCM data that has passed through the digital filter into PCM data;
The audio reproducing apparatus according to claim 1, further comprising: A noise determination unit for determining whether the PCM data from the second decoder is noise;
A reset unit that resets the previous value held by the second decoder when the noise determination unit determines that the noise is;
The sound reproducing device according to claim 1, further comprising: Further comprising a cut-off frequency changing unit for changing the cut-off frequency according to the noise judgment of the noise judgment unit,
4. The audio reproduction device according to claim 3, wherein the cut-off frequency is set lower during the determination as noise than when it is not determined as noise.   The audio reproduction device according to claim 3, further comprising an attenuation unit that attenuates the PCM data to 1/8 when the noise determination unit determines that the noise is noise. A data receiving step of receiving encoded audio data encoded using an audio codec by ADPCM;
A data determination step for determining whether the encoded audio data is low-speed data of 16 kbps or high-speed data other than 16 kbps and higher than 16 kbps;
A first decoding step for decoding the encoded audio data into PCM data when the determination in the data determination step is high-speed data, and a second decoding for loading the encoded audio data and decoding into PCM data when the data determination step is low-speed data A digital filter step for suppressing at least a high frequency of the PCM data decoded in the second decoding step;
An audio conversion step of converting the PCM data into an analog audio signal;
An audio reproduction step of reproducing the analog audio signal;
An audio reproduction method comprising:

Images