![语音/音频处理学术速递[7.28]](https://api.speechhome.com/Resources/Thumb/2025-07-28/1949764371265753088.jpg)
声浪
![语音/音频处理学术速递[7.25]](https://api.speechhome.com/Resources/Thumb/2025-07-25/1948652077144215552.jpg)
![语音/音频处理学术速递[7.24]](https://api.speechhome.com/Resources/Thumb/2025-07-24/1948308650300936192.jpg)
![语音/音频处理学术速递[7.23]](https://api.speechhome.com/Resources/Thumb/2025-07-23/1947919645776416768.jpg)
![语音/音频处理学术速递[7.22]](https://api.speechhome.com/Resources/Thumb/2025-07-22/1947589847946498048.jpg)
![语音/音频处理学术速递[7.21]](https://api.speechhome.com/Resources/Thumb/2025-07-21/1947199836927430656.jpg)
![语音/音频处理学术速递[7.18]](https://api.speechhome.com/Resources/Thumb/2025-07-18/1946108090764103680.jpg)
![语音/音频处理学术速递[7.17]](https://api.speechhome.com/Resources/Thumb/2025-07-18/1946094394151145472.jpg)
![语音/音频处理学术速递[7.16]](https://api.speechhome.com/Resources/Thumb/2025-07-16/1945419716680093696.jpg)
Interspeech2025｜利用动态词表预测与激活的上下文语音识别
近年来，神经网络偏置（Deep Biasing）方法在上下文语音识别（Contextual ASR）任务中取得了显著进展，其旨在提升模型对实体名称、技术术语等罕见短语的识别能力。当前主流方法多基于浅融合或神经网络偏置技术，将上下文短语中的子…
37 0 0
近年来，神经网络偏置（Deep Biasing）方法在上下文语音识别（Contextual ASR）任务中取得了显著进展，其旨在提升模型对实体名称、技术术语等罕见短语的识别能力。当前主流方法多基于浅融合或神经网络偏置技术，将上下文短语中的子…