请问采用PaddleSpeech怎么实现实时语音识别？ #1188
-
|
比如，针对电话的应用场景，对通话过程中做同声翻译，需要进行语音流的实时识别，应该怎么调用？ |
Beta Was this translation helpful? Give feedback.
Replies: 2 comments 2 replies
-
|
先有电话的ASR。已DS2模型为例，流式部分参考https://github.com/PaddlePaddle/PaddleSpeech/blob/develop/paddlespeech/s2t/exps/deepspeech2/model.py#L402 |
Beta Was this translation helpful? Give feedback.
-
|
语音流的断句需要在应用层自己判断吗？由应用层断句后生成一句话的语音文件，再调用deepspeech2的接口进行识别？ |
Beta Was this translation helpful? Give feedback.
-
|
感谢 |
Beta Was this translation helpful? Give feedback.
-
|
目前已经在 server 目录下实现了ws的流式asr，效果还在调试中。可以先体验下。 |
Beta Was this translation helpful? Give feedback.
目前已经在 server 目录下实现了ws的流式asr，效果还在调试中。可以先体验下。