Speech recognition method based on intelligent glasses, intelligent glasses and storage medium

• Main Authors: WEN ZENGQI, ZHANG MINGHUI, ZHANG BAOFENG
• Format: Patent
• Language: Chinese; English
• Published: 25.04.2023
• Subjects: ACOUSTICS; CONTACT LENSES; MUSICAL INSTRUMENTS; OPTICS; PHYSICS; SPECTACLES; SPEECH ANALYSIS OR SYNTHESIS; SPEECH OR AUDIO CODING OR DECODING; SPEECH OR VOICE PROCESSING; SPEECH RECOGNITION; SUNGLASSES OR GOGGLES INSOFAR AS THEY HAVE THE SAME FEATURESAS SPECTACLES

The invention discloses a voice recognition method based on intelligent glasses, the intelligent glasses and a storage medium, and the method comprises the steps: obtaining a user sound signal, and carrying out the semantic recognition of the user sound signal, and obtaining audio semantic information; detecting mouth shape change characteristics of the user, inputting the mouth shape change characteristics into a convergent lip language recognition model, and recognizing to obtain lip language semantic information of the user; and determining target semantic information according to the audio semantic information and the lip language semantic information. The voice recognition accuracy of the intelligent glasses in a complex environment can be improved. 本申请公开了一种基于智能眼镜的语音识别方法、智能眼镜和存储介质，所述基于智能眼镜的语音识别方法包括：获取用户声音信号，将用户声音信号进行语义识别得到音频语义信息；检测用户的口型变化特征，将口型变化特征输入至收敛的唇语识别模型中，识别得到用户的唇语语义信息；根据音频语义信息和唇语语义信息，确定目标语义信息。本申请能提高智能眼镜在复杂环境中语音识别的准确率。