Voice data labeling method and system, electronic equipment and storage medium

Abstract

The invention discloses a voice data labeling method and system, electronic equipment and a storage medium, and the method comprises the steps: firstly screening original voice data, and carrying outthe reading text matching of screened voice, and obtaining proofreading voice and proofreading text; performing word segmentation on the proofreading text to obtain a word segmentation text; performing noise reduction on the proofreading voice to obtain noise reduction voice, and inputting the voice features after feature extraction into a VAD model to obtain VAD effective voice duration of the noise reduction voice; carrying out voice forced alignment on the word segmentation text by adopting an acoustic model to obtain the word-level alignment time, word-level time intervals, segmented texts, the segmented text starting time, the ending time and the text alignment time; determining a speech speed, an effective time ratio and an error word number according to the plurality of times, and performing speech quality inspection; and segmenting the original voice according to the starting time and the ending time of the segmented text, and taking the segmented text and the segmented voice as voice annotation results. The voice annotation text with qualified quality can be automatically acquired.

Description

technical field [0001] The invention relates to the field of voice data processing, in particular to a voice data labeling method, system, electronic equipment and storage medium. Background technique [0002] With the rapid development of speech technology, the demand for reliable and high-quality speech annotation data required for model training is increasing. Especially in the field of speech recognition, it is difficult to obtain a large amount of reliable annotation data in a short time to quickly build a model. The demand for voice annotation data has four characteristics: large data volume, high annotation quality, multiple scenarios, and multiple languages. The traditional purely manual voice data annotation method is difficult to meet the current voice production needs. Therefore, how to automatically obtain voice-annotated text and ensure the quality of the voice-annotated text has become an urgent problem to be solved. Contents of the invention [0003] In vie...

AI Technical Summary

Problems solved by technology

The demand for voice annotation data has four characteristics: large data volume, high annotation quality, multiple scenarios, and multiple languages. The traditional purely manual voice data annotation method is difficult to meet the current voice production needs.

Images