JPH0332230A - Voice coding/decoding system - Google Patents

Abstract

PURPOSE:To realize high efficient coding/decoding by applying multiplexing to a redundant bit to an error detection of a prediction coefficient only at an encoder side, sending the result and avoiding correction filtering at the decoder side when the error in the prediction coefficient is detected. CONSTITUTION:Only error detection 2 detecting whether or not any error takes place is applied at a sender side to a high efficiency coding output without applying error correction. When the code is sent while a redundant bit RB is added, the receiver side extracts the redundant bit RB to apply decoding 4 for error detection. When no error detection is made and no error exists in the prediction coefficient, a prediction coefficient in the high efficient decoding 5 is used to apply correction filtering 6 to the decoded output, the voice characteristic of the auxiliary information SI is emphasized to apply any error correction to main information MI. When the error detection of the prediction coefficient is applied, the correction filtering 6 is controlled to be off. Thus, the system immune to the transmission line error is realized without sacrificing the efficiency by the high efficient coding/decoding system.

Description

[Detailed Description of the Invention] [Summary] Encoding and decoding system for dealing with transmission path errors in digital mobile radio communication systems etc. Transmission without sacrificing efficiency using a high-efficiency code/decoding system The aim is to realize a system that is resistant to path errors.The transmitting side performs error detection on the prediction coefficients in the auxiliary information using high-efficiency coding and adds redundant bits to the transmission information focus, and the receiving side Error detection decoding is performed on the prediction coefficient from the redundant bits, and when no error detection is performed, corrective filtering of vocal tract characteristics is performed on the high efficiency decoding output using the prediction coefficient in high efficiency decoding, and error detection is performed. The correction filtering is configured to stop when the correction filtering is performed.

[Industrial application field]

The present invention relates to a voice encoding/decoding system, and particularly to an encoding and decoding system for dealing with transmission path errors in digital mobile radio communication systems and the like.

In recent years, the full-scale introduction of digital mobile radio (vehicle radio) communication systems has been considered, but since the transmission path is in free space, there are many factors that hinder accurate transmission, and information is often transmitted during transmission. The contents change. Therefore, there is a need for an encoding/decoding system that can cope with this transmission path error.

[Conventional technology]

As shown in Fig. 6, the conventional audio encoding/decoding system performs high-efficiency encoding (1) on a digitized audio signal.
Error correction encoding (2) is performed on the information focus rB consisting of main information as sound source information obtained by encoding the residual signal and auxiliary information such as prediction coefficients (LPG coefficients) and pitch periods.

In this error correction encoding (2), for example, a parity check is performed on the entire information pick I-IB, a redundant pick (RB) is added for error correction, and the information is sent to the transmission line 3.

On the receiving side, the redundant bit-RB from the transmission path 3 is extracted, and the error bit indicated by this redundant bit-RB is extracted! −
EB is subjected to error correction decoding (/10) and then high efficiency decoding (5).

In this way, redundant bits are added to the information bits for protection when sending them to the transmission path 3, thereby providing resistance to transmission path errors.

[Problem to be solved by the invention]

This conventional method has the following problems.

■When the condition of the transmission path was worse than expected, error correction became impossible, or the correct focus was incorrectly corrected.

■Redundant bits were added to both the main information and the auxiliary information in the information bits for transmission, but this redundancy is proportional to the error correction ability and provides a strong error correction ability. This requires a large number of redundant bits, which deteriorates the efficiency of high-efficiency coding.

Therefore, it is an object of the present invention to realize a system that is resistant to transmission path errors without sacrificing the efficiency of a high-efficiency code/decoding system.

[Means to solve the problem]

In order to achieve the above object, in the speech encoding/decoding system according to the present invention, as shown in principle in FIG.
Error detection is performed on the prediction coefficients in the auxiliary information SI using high-efficiency encoding (1), redundant bit 1-RB is added to transmission information bit I, and on the receiving side, the redundant bit 1-RB is Error detection decoding (4) of the prediction coefficients is performed, and when no error is detected, the prediction coefficients in high efficiency decoding (5) are used to perform correction filtering (6) of vocal tract characteristics on the high efficiency decoding (5) output. ), and when an error is detected, the correction filtering (6) is stopped.

[For production]

In the present invention, on the transmitting side, only error detection (2) for detecting whether or not an error has occurred is performed instead of error correction on the high-efficiency encoded output. As a result, the redundant bit RB indicating the result of error detection (2) is smaller than the error correction bit I.
It's just a matter of numbers.

In addition, the redundant signal (RB) is the main information M that encoded the residual signal.
7) Error detection for only the prediction coefficients in the auxiliary information Sl, not for the entire IB (2)
), the number of bits is even smaller (see FIG. 2).

In this way, when transmitting by adding redundant PIP (RB),
On the receiving side, the redundant RB is extracted and error detection decoding (4) is performed.

As a result of reducing the number of redundant bits (RB) on the transmitting side, the receiving side only knows whether there is an error caused by the redundant bits RB.

Therefore, when no error is detected and there is no error in the prediction coefficients, corrective filtering (6) is performed on the decoded output using the prediction coefficients from high-efficiency decoding (5). As shown in , the vocal tract characteristics of the auxiliary information SI are emphasized to correct any errors in the main information Ml.

However, when an error is detected in the prediction coefficient, the correction filtering (6) will emphasize the error, so the correction filtering (6) is turned off. 4 is an explanatory diagram of the correction filtering operation, FIG. 5 is a block diagram showing an embodiment of the present invention, and FIG. 6 is a principle diagram for explaining a conventional example.

In Figure 1, 1... High efficiency encoding, 2...Error detection encoding, 3...transmission line, 4...Error detection decoding, 5...High efficiency decoding, 6...Correction filtering.

In the figures, the same reference numerals indicate the same or corresponding parts.

Claims (1)

[Claims] On the transmitting side, error detection is performed on the prediction coefficients in the auxiliary information by high-efficiency encoding, and redundant bits are added to the transmission information bits, and on the receiving side, the prediction coefficients are extracted from the redundant bits. Error detection decoding is performed, and when no error is detected, correction filtering of the vocal tract characteristics is performed on the high efficiency decoding output using the prediction coefficients from high efficiency decoding, and when an error is detected, the correction filtering is performed. A speech encoding/decoding method characterized by stopping the .