The invention discloses a voice wake-up method and device, a vehicle, equipment and a storage medium. The method comprises the following steps: receiving voice information of a user; inputting acoustic characterization in the voice information and keyword characterization in the predefined keywords into a pre-trained model to obtain fusion characterization, inputting the fusion characterization into a decoder, and outputting the posterior probability of each frame of modeling unit; and decoding the posterior probability of each frame of modeling unit based on the decoded image to obtain a decoding result, comparing an end point score with a preset score threshold if the decoding result is decoding to an end point, and triggering a wake-up system when the end point score is greater than the preset score threshold. According to the method, the voice system can be awakened by using the self-defined keywords, the problem that a user needs to memorize preset keywords to perform voice interaction is solved, and the awakening rate and the false awakening rate are ensured by enhancing the keywords and setting the awakening conditions through the keyword enhancing module.
本申请公开了一种语音唤醒方法、装置、车辆、设备以及存储介质,所述方法包括:接收用户的语音信息;将语音信息中的声学表征和预定义关键词中的关键词表征输入到预先训练的模型中得到融合表征,将融合表征输入到解码器,输出每一帧建模单元的后验概率;基于解码图对每一帧建模单元的后验概率进行解码处理,得到解码结果,若解码结果为解码到终点,则将终点分数与预设分数阈值进行比较,当终点分数大于预设分数阈值时,则触发唤醒系统。本申请能够利用自定义关键词对语音系统进行唤醒,解决了用户需要记忆预先设置好的关键词来进行语音交互的问题,并且通过关键词增强模块对关键词的增强和唤醒条件的设定,保证了唤醒率和误唤醒率。
Voice wake-up method and device, vehicle, equipment and storage medium
一种语音唤醒方法、装置、车辆、设备以及存储介质
2025-05-30
Patent
Electronic Resource
Chinese
Voice wake-up method and device, equipment and storage medium
European Patent Office | 2021
|Voice wake-up method, related device, equipment, system and storage medium
European Patent Office | 2025
|Vehicle voice control method, device and equipment, vehicle and storage medium
European Patent Office | 2020
|Vehicle wake-up detection method and device, computer equipment and storage medium
European Patent Office | 2024
|