Method, device, equipment and storage medium for short-term speech signal processing

Abstract

The embodiment of the invention discloses a short-term speech signal processing method, device, equipment and storage medium. Among them, the method includes: acquiring the near-end time domain signal, and determining the far-end time-domain signal and the error time-domain signal matching the near-end time-domain signal; determining the audio collection state matching the near-end time-domain signal, the audio collection state Including: single-speak state or double-speak state; determine the residual echo amplitude spectrum and environmental noise amplitude spectrum corresponding to the near-end time-domain signal according to the far-end time-domain signal, error time-domain signal and audio acquisition status; according to the residual echo amplitude spectrum, Ambient noise magnitude spectrum and error time domain signal to generate an output time domain signal that matches the near-end time domain signal. The technical solution of the embodiment of the present invention can effectively suppress the residual echo and environmental noise in the voice signal in an echo scene, and improve the clarity of the voice signal.

Description

technical field [0001] Embodiments of the present invention relate to audio processing technologies, and in particular to a short-term speech signal processing method, device, equipment, and storage medium. Background technique [0002] With the continuous development of terminals, more and more terminals have audio input and audio output functions, and the output audio is picked up by the audio input device again, forming an echo. For example, a smart device with a speaker and a microphone. The presence of an echo signal will affect the quality of the audio signal. [0003] In the prior art, the processing of the echo of the terminal generally adopts an adaptive filter to construct an echo canceller to cancel the echo. The adaptive filter is subtracted from the near-end audio signal picked up by the microphone to output an estimated echo signal, and the subtraction result is called an error signal. Ideally, the error signal is considered to be the effective speech signal...

AI Technical Summary

Problems solved by technology

[0004] The existing technology has the following defects: when the terminal uses the audio input and audio output functions at the same time, for example, when the speaker and the microphone of the smart device work at the same time, the echo signal in the preprocessed sound signal is not eliminated cleanly, and still contains residual Echo and ambient noise

In the short-term voice signal processing system of the terminal, the residual echo and environmental noise in the short-term voice signal will reduce the clarity of the voice signal and affect the normal operation of the system

For example, in the voice message application scenario, residual echo and environmental noise will affect the voice quality; for a speech recognition system with a small word size, residual echo and environmental noise will affect the recognition rate

Images