Voice encoding method

Abstract

Provided is a voice encoding method. The voice encoding method mainly comprises two modules, namely encoders and decoders. High-frequency signals of input voice, after being processed via a high-frequency filter and a high-frequency encoder, are transmitted to a demultiplexer, while low-frequency signals of the input voice, after being processed via a low-frequency filter and a narrowband encoder, are transmitted to the demultiplexer. The signals, after passing through the demultiplexer, are transmitted to a multiplexer on the receiver side via a channel. The high-frequency signals output from the multiplexer are decoded by the high-frequency decoder and then are output; and the low-frequency signals, after being decoded by the narrowband decoder, are output together in integration with high and low frequency signals. The voice encoding method has a sampling rate of 6 kHz and a transmission code rate of 6.4 to 25.85 kbit / s. According to the voice encoding method, improvements are made on aspects of the removal of background noises and PLC processing, and thus significant improvement is made on the performance of the algorithm.

Description

technical field [0001] The invention relates to the field of software coding, in particular to a voice coding method. Background technique [0002] Speech is the most convenient, effective and common way for people to convey information. It can not only convey semantic information, but also convey emotions, attitudes and other personal information. Due to the important position of speech in human communication, its analysis and application have also become one of the important research directions in the field of information and signal processing. At present, speech processing mainly includes the following fields: speech coding, speech enhancement, speech recognition and speaker recognition, and speech synthesis. Therefore, it is of great significance to conduct in-depth and multi-angle analysis of speech and its models, obtain speech characteristics and carry out further processing. With the rapid development of the Internet, a new Internet-based telephone service - VoIP s...

AI Technical Summary

Problems solved by technology

Most of the speech codec systems in use today are based on the existing telephony bandwidth narrowband speech, which is usually limited to about 300Hz-340Hz with a sampling rate of 8kHz. This inherent bandwidth of the traditional public switched telephone network (PSTN) Restrictions restrict the further improvement of voice communication quality, resulting in unsatisfactory voice naturalness, music processing and some special processing

Although this will not seriously reduce the subjective quality of voice, in many other applications, such as entertainment, voice broadcasting, teleconferencing, multi-coal communication and high-definition television, etc., higher quality voice is urgently needed

Images