SpeechRecognizer 在第一次收听时抛出 onError

SpeechRecognizer throws onError on the first listening

在Android5 我遇到了奇怪的问题。第一次调用 SpeechRecognizer 的 startListening 会导致错误代码为 7 (ERROR_NO_MATCH) 的 onError。

我使用以下代码制作了测试应用程序:

if (speechRecognizer == null) {

    speechRecognizer = SpeechRecognizer.createSpeechRecognizer(this);
    speechRecognizer.setRecognitionListener(new RecognitionListener() {
        @Override
        public void onReadyForSpeech(Bundle bundle) {
            Log.d(TAG, "onReadyForSpeech");
        }

        @Override
        public void onBeginningOfSpeech() {
            Log.d(TAG, "onBeginningOfSpeech");
        }

        @Override
        public void onRmsChanged(float v) {
            Log.d(TAG, "onRmsChanged");
        }

        @Override
        public void onBufferReceived(byte[] bytes) {
            Log.d(TAG, "onBufferReceived");
        }

        @Override
        public void onEndOfSpeech() {
            Log.d(TAG, "onEndOfSpeech");
        }

        @Override
        public void onError(int i) {
            Log.d(TAG, "onError " + i);
        }

        @Override
        public void onResults(Bundle bundle) {
            Log.d(TAG, "onResults");
        }

        @Override
        public void onPartialResults(Bundle bundle) {
            Log.d(TAG, "onPartialResults");
        }

        @Override
        public void onEvent(int i, Bundle bundle) {
            Log.d(TAG, "onEvent");
        }
    });
}

final Intent sttIntent = new Intent(RecognizerIntent.ACTION_RECOGNIZE_SPEECH);
sttIntent.putExtra(RecognizerIntent.EXTRA_LANGUAGE_MODEL,
        RecognizerIntent.LANGUAGE_MODEL_FREE_FORM);
sttIntent.putExtra(RecognizerIntent.EXTRA_LANGUAGE, "en");
sttIntent.putExtra(RecognizerIntent.EXTRA_LANGUAGE_PREFERENCE, "en");

speechRecognizer.startListening(sttIntent);

并在第一次 startListening 调用后有此日志消息:

onError 7
onReadyForSpeech
onBeginningOfSpeech
onEndOfSpeech
onResults

并且以下消息接连不断 startListening 调用:

onRmsChanged
...
onRmsChanged
onReadyForSpeech
onRmsChanged
...
onRmsChanged
onBeginningOfSpeech
onRmsChanged
...
onRmsChanged
onEndOfSpeech
onRmsChanged
onRmsChanged
onRmsChanged
onResults

那么,这个错误的原因是什么,我该如何解决?

我遇到了同样的问题,但我找不到解决方法,所以如果 startListening 和 onError 之间的时间过短,我最终只是在 onError 中调用 return。

protected long mSpeechRecognizerStartListeningTime = 0;

protected synchronized void speechRecognizerStartListening(Intent intent) {
    if (mSpeechRecognizer != null) {
        this.mSpeechRecognizerStartListeningTime = System.currentTimeMillis();
        RLog.d(this, "speechRecognizerStartListening");
        this.mSpeechRecognizer.startListening(intent);
    }
}
...
@Override
public synchronized void onError(int error) {
    RLog.i(this, this.hashCode() + " - onError:" + error);

// Sometime onError will get called after onResults so we keep a boolean to ignore error also
    if (mSuccess) {
        RLog.w(this, "Already success, ignoring error");
        return;
    }

    long duration = System.currentTimeMillis() - mSpeechRecognizerStartListeningTime;
    if (duration < 500 && error == SpeechRecognizer.ERROR_NO_MATCH) {
        RLog.w(this, "Doesn't seem like the system tried to listen at all. duration = " + duration + "ms. This might be a bug with onError and startListening methods of SpeechRecognizer");
        RLog.w(this, "Going to ignore the error");
        return;
    }

// -- actual error handing code goes here.
}

一旦您为每个屏幕配置 "Okay Google" 功能,就会出现错误。

原来是这个原因!

关闭该功能,问题应该就解决了

我在多台设备上都遇到了同样的问题。似乎 onError(7) 总是在 onReadyForSpeech() 之前被调用,所以如果要避免使用丑陋的时间,你可以这样做:

public void start(){
    performingSpeechSetup = true;
    speechRecognizer.startListening(intent);
}

并在 RecognitionListener 中:

public void onReadyForSpeech(Bundle bundle) {
    performingSpeechSetup = false;
}

@Override
public void onError(int error) {
    if (performingSpeechSetup && error == SpeechRecognizer.ERROR_NO_MATCH) return;
    // else handle error
}

完成了一种解决方法。

这是常规流程

onReadyForSpeech -->onBeginningOfSpeech-->onEndOfSpeech -->onResults

但是流量很奇怪

onError(不匹配)-->onReadyForSpeech-->onBeginningOfSpeech-->onEndOfSpeech-->onResults

因此将语音结束时的布尔值设置为真。并检查 onError 以确保它在演讲结束后抛出错误!

speech.startListening(recognizerIntent);
isEndOfSpeech = false;

 @Override
    public void onError(int error) {
        if (!isEndOfSpeech)
            return;
}


 @Override
    public void onEndOfSpeech() { 
        isEndOfSpeech = true; 
    }

结果证明对我来说非常简单。语音识别的启动声音太大,一开始就触发了收听过程。调低系统声音会有帮助。 (音量键)