JPS62285541A - Error control method in voice digital transmission of vocoder system - Google Patents

Abstract

PURPOSE:To attain decoding processing without defect by adding a parity bit at each frame of a transmission data and allowing the reception decoding side to detect an error. CONSTITUTION:A voice input is sectioned at each prescribed time, a parameter characterizing the voice waveform in each section is analyzed and the result is transmitted with digital coding. A sender side coding circuit adds a parity bit at each frame of each analysis data, transmits the result, and a reception side decoding circuit detects the presence of an error at each frame of the reception data and if any error is detected, the data of the frame is replaced by the data of the frame just before to apply decoding processing. That is, a parity generation circuit 16 adding a parity bit for the error detection is added to an output of a multiplexer 6 in the coding circuit and a detection circuit 17 and a register 18 are added to the decoding circuit. Thus, the deterioration of the quality of the voice decoded due to transmission error is reduced.

Description

[Detailed Description of the Invention] 3. Detailed Description of the Invention (Field of Industrial Application) It is well known that there is a VOCODER system as a coding system that digitally encodes and transmits audio at a low bit rate. . In this method, the audio is
It is divided into sections of 20m5ec, and the parameters that characterize the waveform in each section are analyzed and transmitted.The results of determining whether the sound is voiced or unvoiced, and in the case of voiced sound, its pitch frequency and spectrum information are transmitted. It is customary to express the information using various parameters, 1, and digitally encode it before transmitting it. The present invention relates to means for reducing the influence of transmission errors that occur in such vocoder transmission on the sound quality of restored speech.

(Prior art) Figure 4 is a side view of the configuration of a conventional encoding circuit using a vocoder system.
FIG. 5 is a side view of the configuration of a conventional decoding circuit, and both figures show an example of analyzing and synthesizing speech using digital arithmetic processing. The symbols used in all the figures below are common, and the same symbols represent equipment with the same function.

In FIG. 4, l is an AD changer, 2 is a spectrum information analysis circuit, 3 is a spectrum information encoder, 4 is a sound source information analysis circuit that detects the pitch frequency and determines whether the sound is voiced or unvoiced, 5 is a sound source information encoder, 6 is a multiplexer that multiplexes the outputs of both encoders 3 and 5 to form transmission data, 7 is a synchronization signal generation circuit, 8 is a switch, and 9 is a sequence of this encoding circuit. This is a control circuit that controls the

After the audio input is digitally encoded by the AD converter 1,
The spectrum analysis circuit 2 and the sound source information analysis circuit 4 separate and analyze each analysis time. The analysis time is set to 5 to 20 m5 depending on the transmission speed and the like. The analyzed results are quantized into transmission data by encoders 3 and 5, respectively, and multiplexed by multiplexer 6. (Many documents are known for explaining such vocoder methods.) In the vocoder method, one section of analysis data is generally a code of several tens of bits, and the decoder does not know the frame position of the code. It is necessary to be present. Therefore, it is necessary to synchronize by transmitting a synchronization signal at the start of data transmission.A synchronization signal generator is a circuit that generates the synchronization signal, and the output is switched to the output of the multiplexer 6 by a switch 8.

FIG. 6 is an example of the format of data transmitted by the circuit of FIG. 4. In the figure, 5YNC is a synchronization signal, and D is one frame (bit! (N)) of analysis data. First, a synchronization signal is transmitted, and then analysis data is transmitted.A PN code based on an M-sequence code or the like and having good autocorrelation characteristics is used for the synchronization signal.

The decoding circuit shown in FIG. 5 will be explained. In the figure, 10 is a synchronization signal detection circuit, 11 is a control circuit that controls the voice synthesis operation, 12 is a register for receiving data, 13 is a pulse generation circuit for the sound source, and 14 is a filter that generates voice according to the spectrum information received. Synthesize the spectrum. 15 is an output DA converter. In signal reception by this decoding circuit, first, the synchronization signal detection circuit 10 detects the first synchronization signal 5YNC from the reception input, and the signal is passed through the control circuit 11 to manage the frame position of the received data. In speech synthesis, received input is stored in the register 12 frame by frame, separated into sound source information and spectrum information, and decoded. The decoded sound source information controls the pulse generation circuit 13 to generate impulse trains according to voiced sounds, unvoiced sounds, and pitch frequencies. On the other hand, the spectrum information enters the filter 14 from the register 12, the coefficients of the filter 14 are controlled so that the spectrum is the same as the original voice, the impulse train is converted into a p-wave, the voice is composed, and the analog waveform is converted into an analog waveform by the DA converter 15. It is converted to and output.

(Specific Object of the Invention) In conventional methods that do not perform error control, if an error occurs during transmission, incorrect data will be used in decoding as the data used to generate the sound source or the data that determines the characteristics of the filter. Depending on the position of the bit where the error occurred, the pitch frequency or filter characteristics may change significantly, which not only reduces the clarity and naturalness of the restored voice, but also changes the sound to an abnormal sound that is different from the voice. This may cause the restored sound to be difficult to hear. The present invention adds a parity bit to each frame of transmission data before transmitting it, performs error detection on the reception/decoding side, and if an error is detected, synthesizes audio using the data received in the previous frame. By doing so, the objective is to perform decoding processing without the above-mentioned drawbacks.

(Structure of the Invention) The bit structure of one frame of transmission data according to the invention is shown in FIG. SY in this figure is a synchronization bit, DATA
is analysis data, P is a parity pit for error detection, and 1 bit (N K 1) is assigned to each bit. N is the number of bits in one frame.

The synchronization bit is used by the decoding circuit to monitor the synchronization state and the presence or absence of reception input.

FIG. 1 is a block diagram of an encoding circuit embodying the present invention, and FIG. 2 is a block diagram of a decoding circuit of the same embodiment.

First, in Figure 1, 1 to 9 are common to Figure 4, and 1 is AD.
i converter, 2 is a spectrum analysis circuit, 3 is an encoder, 4 is a sound source analysis circuit, 5 is an encoder, 6 is a multiplexer, 7 is a synchronization signal generation circuit, 8 is a switch, 9 is a control circuit, A parity generation circuit 16 is added to the output of the multiplexer 6 for adding a parity bit for error detection. The output of the multiplexer 6 has the data format shown in FIG.
A parity bit is added in , resulting in the form shown in Figure 3.

Next, FIG. 2 will be explained. 10.11゜1 in this figure
3 to 15 are common to FIG. 5, and 10 is a synchronization signal detection circuit;
11 is a control circuit, 13 is a pulse generation circuit, 14 is a filter, and 15 is a DA converter, which also includes an error detection circuit 17 that detects errors in received data for each frame and a register that is also controlled by this error detection output. 18 have been added.

The register 18 accumulates input data in units of one frame,
Speech synthesis parameters are supplied to the pulse generation circuit 13 and the filter 14, but if an error is detected by the error detection circuit 17, the received data of the previous frame is replaced and supplied as the speech synthesis parameters.

(Operation of the Invention) Here, the basis on which the present invention is possible will be explained. Audio is a signal with extremely high redundancy, and the spectrum and pitch frequency change slowly compared to the time of one frame, and there are few changes between frames. Therefore, even if audio is synthesized by replacing the data of the frame in which the error was detected with the data of the previous frame, there is little change in sound quality. On the other hand, if received data containing errors is directly synthesized into speech, the above-mentioned problem occurs. In other words, the quality of the voice restored by replacing it with the data of the previous frame is better.

Next, there is a method of using error correction codes as a countermeasure against transmission errors, but there is a problem that error correction requires a large number of bits of parity bits and a wide transmission bandwidth. be. Furthermore, although the number of focus points of the parity bit for error detection may be, for example, 1 bit for 1-bit error detection, it is well known that more reliable error detection can be achieved by using a cyclic code or the like. The actual number of parity bits is determined in consideration of the error rate distribution expected to occur in the line used, the length of burst errors, etc.

(Effects of the Invention) By implementing the present invention, when applying vocoder digital transmission to a wireless line, it is possible to reduce deterioration in the quality of restored voice due to transmission errors.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a vocoder encoding circuit implementing the present invention, FIG. 2 is a block diagram of a decoding circuit implementing the present invention,
Fig. 3 is a bit configuration diagram of one frame of transmission data in the present invention, Fig. 4 is a configuration diagram of a conventional vocoder encoding circuit, Fig. 5 is a configuration diagram of a conventional decoding circuit, and Fig. 6 is a conventional This is the format of the transmitted data. DESCRIPTION OF SYMBOLS 1... AD conversion circuit, 2... Spectrum information analysis circuit, 3... Spectrum information encoder, 4...
Sound source information analysis circuit, 5... Sound source information encoder, 6.
...Multiplexer, 7...Synchronization signal generation circuit, 8.
...Switcher, 9...Control circuit, 10...Synchronization signal detection circuit, 11...Control circuit, 12...Register,
13... Pulse generation circuit of sound source, 14... Filter, 15... DA converter, 16... Parity generation circuit, 17... Error detection circuit, 18... Register.

Claims (1)

[Claims] In a vocoder-based audio encoding and decoding circuit that divides audio input into fixed time intervals, analyzes parameters that characterize the audio waveform within each segment, digitally encodes the data, and transmits the data, the transmitting side encoding circuit A parity bit is added to each frame of each analysis data and transmitted, and the receiving side decoding circuit detects the presence or absence of an error in each frame of received data. If an error is detected, the data of that frame is An error control method in vocoder-based audio digital transmission characterized by performing decoding processing by replacing frame data with frame data.