You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
------------------ 原始邮件 ------------------
发件人: "Wei ***@***.***>;
发送时间: 2024年3月4日(星期一) 晚上7:18
收件人: ***@***.***>;
抄送: ***@***.***>; ***@***.***>;
主题: Re: [k2-fsa/sherpa-ncnn] 长热词导致端点检测失效的NCNN流式语音识别模型问题 (Issue #318)
同 #281 (comment)
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you authored the thread.Message ID: ***@***.***>
较长热词(比如四个字及以上)影响ncnn流式语音识别模型正常工作,比如python-api-examples/speech-recognition-from-microphone-with-endpoint-detection.py中添加hotwords_file=r"\sherpa-ncnn-streaming-zipformer-bilingual-zh-en-2023-02-13\hotwords_mix.txt",hotwords_score=2.0后,识别到热词时会在当前行序号立马结束识别,仿佛是端点检测失效一般。
The text was updated successfully, but these errors were encountered: