A method for real-time speech separation and speech transcription

Abstract

The present invention is a method for real-time voice separation and voice transcription, comprising: a hardware acquisition module, a voice separation module, and a voice transcription module, wherein the hardware acquisition module is used to collect voice digital signals; The voice digital signals collected in different directions are separated; the voice transcription module transcribes the various voice digital signals separated by the voice separation module into text, and the microphone pick-up module of the hardware acquisition module collects voice signals, only need to configure the angle Parameters; the voice separation module can effectively separate multi-person conversations by speakers in real time and perform real-time voice transcription; at the same time, the voice separation module can effectively reduce the interference of environmental noise and transcribe the sound source in a fixed direction, so that Effectively separate overlapping dialogue voices and perform effective voice transcription.

Description

technical field [0001] The invention relates to the field of computers, in particular to a method for real-time speech separation and speech transcription. Background technique [0002] In service dialogue scenarios like insurance and bank counters with fixed positions, it is necessary to effectively record the dialogue between the two parties in a slightly noisy environment; [0003] The existing desktop pointing sound pickup products can only pick up the sound of the user, that is, the user can pick up the sound in the near field, and cannot obtain the voice of the person in the conversation at the same time. If there are many people talking in the scene, it is necessary to arrange multiple near-field Pickup equipment, and it is very close to the speaker, usually within 20 cm. This method can judge and separate multiple human voices in different directions in real time through the microphone array pickup placed on the desktop, and output corresponding text information acc...

AI Technical Summary

Problems solved by technology

[0003] The existing desktop pointing sound pickup products can only pick up the sound of the user, that is, the user can pick up the sound in the near field, and cannot obtain the voice of the person in the conversation at the same time. If there are many people talking in the scene, it is necessary to arrange multiple near-field Pickup equipment, and very close to the speaker, usually within 20 cm

Images