SPEECH RECOGNITION AND SPEECH RECOGNITION MODEL TRAINING METHOD AND APPARATUS

摘要

Disclosed are a speech recognition method, and a speech recognition model training method and apparatus. The speech recognition method comprises: recognizing a target word speech from a mixed speech, obtaining an anchor extraction feature of the target word speech based on the target word speech, and taking the anchor extraction feature of the target word speech as an anchor extraction feature of a target speech (100); obtaining a mask code of the target speech according to the anchor extraction feature of the target speech (110); and recognizing the target speech according to the mask code of the target speech (120).