Training method of recognition model, recognition method, electronic device, and storage medium
By processing the text data of wake words using TTS to generate speech samples and training a recognition model using CNN and GRU models, the problem of high cost and low efficiency in wake word detection in existing technologies is solved, achieving low-cost and high-efficiency wake word recognition.
CN114267342BActive Publication Date: 2026-06-12BEIJING BAIDU NETCOM SCI & TECH CO LTD
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BEIJING BAIDU NETCOM SCI & TECH CO LTD
- Filing Date
- 2021-12-21
- Publication Date
- 2026-06-12
Smart Images
Figure CN114267342B_ABST
Abstract
The present disclosure provides a training method of a recognition model, a recognition method, an electronic device and a storage medium, relates to the field of artificial intelligence, and particularly relates to the technical field of speech recognition, deep learning and the like. The specific implementation scheme is as follows: a first speech sample containing a wake-up word, a second speech sample not containing the wake-up word, a first label of the first speech sample, and a second label of the second speech sample are obtained; wherein the first speech sample comprises speech data obtained by performing text-to-speech (TTS) processing on text data containing the wake-up word; a first acoustic feature of the first speech sample and a second acoustic feature of the second speech sample are obtained; and a recognition model is trained by using the first acoustic feature, the second acoustic feature, the first label and the second label. According to the present disclosure, the collection time and cost of sample data can be reduced.
Need to check novelty before this filing date? Find Prior Art