JPH06161497A - Speech decoder and speech encoder - Google Patents

Abstract

PURPOSE:To provide the speech decoder and speech encoder which can perform processing of speech data in plural series on a time-division basis. CONSTITUTION:The speech decoder and speech encoder are equipped with encoded speech data input buffers IBF1 and IBF2, internal state information memories DM1 and DM2 of a VSELP system decoder VD, and decoded speech output buffers OBF1 and OBF2. One-frame data of the encoded speech data in two series are inputted to the input buffers IBF1 and IBF2 respectively and while the data in the input buffer IBF1 are decoded on the basis of the contents of the internal state information memory DM1 and outputted, the contents of the internal state information memory DM1 are rewritten into the processing result; and the data in the input buffer IBF2 are similarly processed on the basis of the contents of the internal state information memory DM2. The speech decoder 4 performs the operation by time-division processing at every frame of speech data in the two series. Further, the speech encoder 2 has similar constitution.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a speech decoder and a speech encoder which can be used in a digital cellular phone such as a digital cellular system and which time-division-processes digital speech data of a plurality of systems.

[0002]

2. Description of the Related Art In recent years, so-called mobile phones such as cellular phones have been digitized. In the past, voice was transmitted by analog modulation, but in these digital mobile phones, analog voice signals are converted into analog signals.
After being converted into PCM by a D converter, the number of bits per time is reduced by a voice encoder such as VSELP system, and then modulated into a radio frequency band by a digital modulation system such as ¼π shift QPSK and transmitted. On the other hand, on the receiving side, the voice is reproduced by performing voice decoding, which is the inverse conversion of voice encoding, from the digital signal obtained by demodulating the received digital modulation signal.

FIG. 6 is a block diagram showing a circuit configuration of a VSELP type digital mobile phone. First, on the transmitting side, the voice input from the microphone MC is processed by the A / D conversion circuit 31.
Becomes quantized / discretized digital voice data by
Information source coding (speech compression) and channel coding (error correction coding) by the VSELP speech coder 32
Is given. Here, in the case of a mobile phone of the full rate digital cellular system, it is converted into encoded voice data having an encoding rate of 11.2 kilobits per second. continue,
The encoded voice data is modulated into a radio frequency by the high frequency section 33 and sent to the base station. On the other hand, on the receiving side,
The modulated signal coming from the base station is demodulated by the high frequency unit 33 and input to the speech decoder 34 as encoded speech data. The voice decoder 34 performs inverse conversion of the above channel coding and information source coding to take out digital voice data, and the D / A conversion circuit 35 converts it into an analog voice signal and outputs the voice from the speaker SP. .

By the way, when using a mobile phone,
It is often used in situations such as outdoors or when moving, and it is inconvenient to take notes about the contents of a call. Therefore, it is possible to store encoded voice data in a mobile phone with a built-in digital memory. it can.

Since a normal telephone conversation consists of two-way voices of the contents transmitted from the user to the other party and the contents transmitted from the other party and received by the user, the digital memory has two codes. It is necessary to store each of the converted voice data. Further, the voice reproduced from the stored encoded voice data also needs to be of two systems simultaneously in order to reproduce the content of the call.

Therefore, two voice decoders are used in order to reproduce the two systems of encoded voice data to be stored at the same time, or in a mobile phone, voice encoding and decoding are usually performed every about 20 milliseconds. It is conceivable that one speech decoder is operated in a time-sharing manner by utilizing the fact that it is divided and processed collectively. If the time-division operation is used, only one speech decoder is enough. However, in the processing for each speech segment, the speech decoder is the internal decoding result of the decoding process up to the previous speech segment. Since the reproduction is performed using the state information, it is not possible to simultaneously reproduce the two systems of encoded audio data by simply performing the time division processing. In addition, VS
In the case of an ELP type speech decoder, the internal state information is the contents of the adaptive codebook, the initial states of the shift registers of the three filters, the LPC coefficient of the previous frame, and the energy value of the previous frame.

Further, due to the diversification and high functionality of communication services in recent years, expansion of multi-channel voice and the like is considered in mobile phones. In this case, for example, if the stereo system is used, a plurality of voice encoders and voice decoders are used to process a plurality of voices, or the time division operation is performed.
A problem similar to that in the case of implementing a system audio decoder occurs.

In order to solve the above problems, the present invention has an object to provide a voice decoder and a voice encoder capable of time-divisionally processing voice data of a plurality of systems.

[0009]

In order to solve the above-mentioned problems, a speech decoder according to the present invention is a speech decoder which decodes encoded speech data of a plurality of systems in a time division manner. The same number of input buffers and the same number of output buffers, and storage means for storing the internal state information of the speech decoder which is the processing result of the previous frame such as the adaptive codebook for each system. The encoded voice data of each system is input to, the encoded voice data in the input buffer and the internal state information of the corresponding system in the storage unit are read, and the code based on the read internal state information Each of the above-mentioned series of processes consisting of: decoding of the encoded audio data, output of the decoded audio data to the output buffer, updating of internal state information of the corresponding system in the storage means, A sequential processing audio decoder for each integration.

Further, the speech coder of the present invention is a speech coder for time-divisionally encoding digital speech data of a plurality of systems, and has the same number of input buffers and the same number of output buffers as the number of systems. A storage means for storing internal state information of a speech coder, which is a result of processing a previous frame such as an adaptive codebook, is provided for each system, and digital audio data of each system is input to the input buffer,
The digital voice data in the input buffer and the internal state information of the corresponding system in the storage means are read, the digital voice data is encoded based on the read internal state information, and the encoded voice is encoded. Outputting data to an output buffer and updating internal state information of a corresponding system in the storage means,
Is a speech coder that sequentially processes the series of processes for each system.

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A voice decoder according to the first invention of the present invention will be described below in accordance with an embodiment in which it is used in a full rate digital cellular digital mobile phone. FIG. 1 is a block diagram showing a circuit configuration of a digital mobile phone for explaining an embodiment of the first invention. On the transmitting side, a voice input from a microphone MC is quantized / discretized by an A / D conversion circuit 1. Converted into digital voice data, and further voice encoder 2
Source-coded (voice compression) and channel-coded (error-correction coding) are performed by the source codec, and in the case of this full-rate digital cellular type mobile phone, coded voice having a coding rate of 11.2 kilobits per second. Converted to data,
The radio frequency is modulated by the high frequency unit 3 and transmitted to the base station. On the receiving side, the modulated signal coming from the base station is
It is demodulated by the high frequency unit 3 and input to the speech decoder 4 as encoded speech data. The voice decoder 4 performs inverse conversion of the above channel coding and information source coding to take out digital voice data, and the D / A conversion circuit 5 converts it into an analog voice signal and outputs the voice from the speaker SP. .

Further, this mobile phone has a memory M for digital data, and a switch SW1 is provided between the voice encoder 2 and the high frequency unit 3 in order to write and read encoded voice data in the memory M. The switching unit SW2 is provided between the voice decoder 4 and the high frequency unit 3, and switching between writing and calling is controlled by the control unit 6. In the memory M, the coded voice data coded by the encoder 2 on the transmitting side and the coded voice data demodulated by the high frequency section 3 on the receiving side are written and read out. Then, by controlling the switches SW1 and SW2, only the voice input from the microphone MC is recorded in the memory M, the recording and the transmission to the base station are simultaneously performed, and the encoded voice received from the base station and demodulated It is possible to perform only various kinds of recording such as recording data and transmitting the recorded contents of the memory M to the base station.

After the end of the call, in order to reproduce the voice recorded in the memory M, the encoded voice data is stored in the memory M.
Read from the D / A and decoded by the decoder 4.
Audio is output via the conversion circuit 5 and the speaker SP.

Here, the encoded voice data recorded in the memory M has two systems, that is, the transmitting side and the receiving side, and needs to be reproduced at the same time in order to understand the contents of the call.

In the full rate digital cellular system, V
A voice encoding method called the SELP method is adopted. The VSELP speech decoder VD is configured as shown in FIG. 2 and has two codebooks, that is, an adaptive codebook 11 and a vector sum codebook 12, and the vector sum of these codebook outputs is connected in three cascades. A voice is generated by passing the connected pitch pre-filter f1, synthesis filter f2, and post filter f3. At this time, the VSELP audio decoder VD
Encoded audio data obtained by encoding audio in the 0-millisecond section (hereinafter referred to as frame) and VSEL that is the processing result of the previous frame
Based on the internal state information of the P-system speech decoder VD, two codebook outputs, gains to be multiplied by the codebook outputs, coefficients of three filters, and initial states of internal shift registers are determined. The internal state information of the VSELP audio decoder VD is, specifically, the contents of the adaptive codebook 11, the initial state of the shift register of the three filters f1 to f3, the LPC coefficient of one frame before, and the energy of one frame before. Is the value of.

As shown in FIG. 3, a speech decoder 4 according to an embodiment of the present invention includes a first encoded speech data input buffer IBF1, a second encoded speech data input buffer IBF2, and a first encoded speech data input buffer IBF2. Decoder Decoder Internal State Information Memory DM
A combination of the first and second decoder internal state information memory DM2, the first decoded audio output buffer OBF1 and the second decoded audio output buffer OBF2 in the VSELP system audio decoder VD is as follows. Works like.

For example, the encoded voice data on the transmitting side and the receiving side recorded in the memory M of FIG.
The case of outputting from P will be described. Each encoded voice data is read alternately for each frame. The transmitter side encoded voice data is accumulated in the first input buffer IBF1,
The receiving side encoded voice data is the second input buffer IBF2.
Accumulated in.

Here, the VSELP speech decoder VD first reads the internal state information of the decoder from the first decoder internal state information memory DM1 and then the adaptive codebook 11
And the initial state of the shift register of the three filters f1 to f3 are set, and the LPC coefficient of the previous frame and the energy of the previous frame are held. Next, the encoded audio data in the first input buffer IBF1 is read, audio decoding is performed, and the decoded audio data is output to the first output buffer OB.
At the same time as writing to F1, the internal state information of the first decoder internal state information memory DM1 is rewritten to this processing result.

Subsequently, the VSELP type speech decoder V
D reads the decoder internal state information from the second decoder internal state information memory DM2 and sets the contents of the adaptive codebook 11 and the initial states of the shift registers of the three filters f1 to f3, The LPC coefficient and the energy of the previous frame are retained. Next, the encoded audio data in the second input buffer IBF2 is read and audio decoding is performed, and the decoded audio data is output to the second output buffer OBF2.
At the same time as the writing, the internal state information of the second decoder internal state information memory DM2 is rewritten with this processing result.

As a result, one frame of voice decoded from the encoded voice data on the transmitting side is accumulated in the first output buffer OBF1, and the encoded voice data on the receiving side is stored in the second output buffer OBF2. One frame of audio decoded from is stored. The decoded audio data in the first output buffer OBF1 and the decoded audio data in the second output buffer OBF2 are added and output, thereby synthesizing the decoded audio on the transmitting side and the receiving side for each frame. You can get the voice. By repeating the above operation, the memory M
The bidirectionally encoded voice data stored in is simultaneously reproduced for each frame.

Next, a description will be given of a speech coder of an embodiment according to the second invention of the present invention. An embodiment of the mobile phone shown in FIG. 4 in which the full rate digital cellular mobile phone shown in FIG. Follow the instructions below.

FIG. 4 is a block diagram showing a circuit configuration of a stereo type mobile phone having two systems of voice input and output, and FIG. 5 is a block diagram showing a circuit configuration of a voice encoder showing one embodiment of the present invention. Is.

As shown in FIG. 4, the operation of the portable telephone having two systems of input / output is such that the transmitting portion, the receiving portion and the high frequency portion of the portable telephone of FIG. Encoding of data and decoding of encoded voice data are performed by a single voice encoder 2 and voice decoder 4, respectively. This speech decoder 4 has been described in FIG.

As shown in FIG. 5, the speech coder 2 includes a first digitized speech data input buffer IBF1 ',
Second digitized voice data input buffer IBF
2 ', first encoder internal state information memory CM1, second
The encoder internal state information memory CM2, the first encoded voice data output buffer OBF1 ′, and the second encoded voice data output buffer OBF2 ′ are combined with the VSELP type voice encoder VC. To work. It should be noted that the configuration of the VSELP system speech coder is for a trader who analyzes and encodes two codebook outputs in the VSELP system decoder, gains to be multiplied in each codebook, and coefficients of three filters. Since it is a well-known technique, it will be omitted.

This operation starts with the two-system microphone MC.
The audio data input from the A / D converter 1 ′ are digitized by the first input buffer IBF.
1'and the second input buffer IBF2 '.

The VSELP type speech coder VC starts from the first coder internal state information memory CM1 to the coder V.
The internal state information of C is read to set the contents of the adaptive codebook and the initial states of the shift registers of the three filters, and the LPC coefficient of the previous frame and the energy of the previous frame are held. Next, the audio data in the first input buffer IBF1 'is read, audio encoding is performed, and the encoded audio data is written in the first output buffer OBF1', and at the same time, the first encoder internal state information memory CM1. The internal state information of the VSELP type speech coder VC stored in (1) is rewritten with this coding result.

Subsequently, the VSELP type speech coder V
C reads the encoder internal information from the second encoder internal state information memory CM2 and sets the contents of the adaptive codebook 11 and the initial states of the shift registers of the three filters f1 to f3, The LPC coefficient and the energy of the previous frame are retained. Then the second input buffer I
The audio data in BF2 'is read, audio encoding is performed, and the encoded audio data is output to the second output buffer OBF2'.
At the same time as the writing, the internal state information of the VSELP type speech encoder VC of the second encoder internal state information memory CM2 is rewritten with this encoding result.

As a result, the first output buffer OBF1 '
The second output buffer OBF2 'and the second output buffer OBF2' are input from the L side and the R side of the microphone MC, respectively, and the encoded encoded audio data is accumulated. First output buffer OBF1 '
The encoded audio data in the second output buffer OBF
By reading out the coded voice data in 2'and inputting them to the high frequency section 3, the voices of two systems for one frame can be coded.

By repeating the above operation, two systems of voice coding can be performed.

Also in speech decoding, in the embodiment of the speech decoder according to the above-mentioned first invention, the two systems of encoded speech data input are not read from the memory, but two systems are inputted from the high frequency section 3. As encoded voice data of
Moreover, the left and right channel sounds to be output are independently output from the two speakers SP, so that two systems of reproduced audio outputs can be obtained for two systems of encoded audio inputs.

Further, in the above embodiment, the input buffer,
By increasing the number of output buffers and internal state information memory, a multi-channel voice decoder and voice encoder can be constructed.

[0032]

According to the present invention, a speech decoder for outputting the same number of decoded speeches to a plurality of systems of encoded speech data input without a large increase in circuit scale and a cost increase, and a plurality of systems. It is possible to realize a speech coder that outputs the same number of encoded speech data as the digital speech data input of

By using the audio decoder according to the present invention, it is possible to simultaneously decode two-way encoded audio data recorded in a memory or the like and reproduce the audio, and the audio according to the present invention. By using the decoder and the voice encoder, it is possible to simultaneously decode and encode the encoded voices of a plurality of channels in a multi-channel digital mobile phone.

[Brief description of drawings]

FIG. 1 is a block diagram showing a circuit configuration of a full-rate digital cellular mobile phone.

FIG. 2 is a block diagram showing a circuit configuration of a VSELP audio decoder.

FIG. 3 is a block diagram showing a circuit configuration of a speech decoder of the present invention.

FIG. 4 is a block diagram showing a circuit configuration of a stereo type mobile phone having two systems of audio input / output.

FIG. 5 is a block diagram showing a circuit configuration of a speech coder of the present invention.

FIG. 6 is a block diagram showing a circuit configuration of a conventional digital mobile phone.

[Explanation of symbols]

2 speech encoder 3 high frequency section 4 speech decoder 11 adaptive codebook 12 vector sum codebook f1 pitch prefilter f2 synthesis filter f3 post filter VD VSELP system speech decoder VC VSELP system speech encoder IBF1, IBF2 IBF1 ', IBF2' Input buffer OBF1, OBF2, OBF1 ', OBF2' Output buffer DM1, DM2, CM1, CM2 Internal state information memory M memory

Claims (2)

[Claims] 1. A voice decoder for decoding coded voice data of a plurality of systems in a time division manner, the number of input buffers and the same number of output buffers as the number of systems, and an adaptive codebook for each system. Storage means for storing internal state information of the speech decoder, which is the result of processing the previous frame such as, and the like, and the encoded speech data of each system is input to the input buffer, and the encoding in the input buffer is performed. Reading of audio data and internal state information of the corresponding system in the storage means, decoding of encoded audio data based on the read internal state information, output of the decoded audio data to an output buffer A speech decoder, wherein the series of processings 1 to 3 consisting of: updating internal state information of the corresponding system in the storage means are sequentially performed for each system. 2. A voice coder for time-divisionally encoding digital voice data of a plurality of systems, the same number of input buffers and the same number of output buffers as the number of systems, and an adaptive codebook for each system. Storage means for storing the internal state information of the speech coder, which is the result of processing the previous frame of 1., the digital speech data of each system is input to the input buffer, and the digital speech data in the input buffer and Reading the internal state information of the corresponding system in the storage means, encoding digital audio data based on the read internal state information, and outputting the encoded encoded audio data to an output buffer,
Updating the internal state information of the corresponding system in the storage means,
A speech coder, which sequentially processes the series of processes (1) to (4) for each system.