azure 如何等待1分钟的发言人的沉默，然后开始识别

js5cn81o 于 12个月前发布在其他

关注(0)|答案(1)|浏览(107)

我希望Azure语音到文本服务在开始识别语音之前至少等待1分钟。或者说话者自己在完成演讲后点击按钮，然后Azure语音到文本服务应该开始从语音中识别文本。
我试过下面的代码，但它不工作

const speechConfig = SpeechConfig.fromSubscription(
      subscriptionKey,
      serviceRegion,
    );
    const audioConfig = AudioConfig.fromDefaultMicrophoneInput();

    speechConfig.setProperty(
      PropertyId.SpeechServiceConnection_EndSilenceTimeoutMs,
      '60000',
    );

    const recognizer = new SpeechRecognizer(speechConfig, audioConfig);

    recognizer.startContinuousRecognitionAsync();

    console.log(new Date().toString());
    
    recognizer.recognizeOnceAsync((result) => {
      if (result.reason === ResultReason.RecognizedSpeech) {
        const recognized = result.text;
        console.log(new Date().toString());
        console.log(recognized);
}}

字符串

Azure

来源：https://stackoverflow.com/questions/77507045/how-to-wait-for-1-minute-for-the-speakers-silence-before-start-recognition

1条答案

按热度按时间

wfauudbj1#

我得到了答案，Speech_SegmentationSilenceTimeoutMs中允许的最大值是5000，
下面是更新后的代码

const speechConfig = SpeechConfig.fromSubscription(
      subscriptionKey,
      serviceRegion,
    );
    const audioConfig = AudioConfig.fromDefaultMicrophoneInput();

    speechConfig.setProperty(
      PropertyId.Speech_SegmentationSilenceTimeoutMs,
      '5000',
    );

    const recognizer = new SpeechRecognizer(speechConfig, audioConfig);
    console.log(new Date().toString());

    recognizer.recognizeOnceAsync((result) => {
      if (result.reason === ResultReason.RecognizedSpeech) {
        const recognized = result.text;
        console.log(new Date().toString());
}
}

字符串

赞(0）回复(0）举报 12个月前

我来回答

azure 如何等待1分钟的发言人的沉默，然后开始识别

1条答案

相关问题

热门标签

最新问答