[TOC]
## 使用麦克风当音听写示例
~~~
//1.创建SpeechRecognizer对象
SpeechRecognizer mIat= SpeechRecognizer.createRecognizer( );
//2.设置听写参数，详见《MSC Reference Manual》SpeechConstant类
mIat.setParameter(SpeechConstant.DOMAIN, "iat");
mIat.setParameter(SpeechConstant.LANGUAGE, "zh_cn");
mIat.setParameter(SpeechConstant.ACCENT, "mandarin ");
//3.开始听写
mIat.startListening(mRecoListener);
//听写监听器
private RecognizerListener mRecoListener = new RecognizerListener(){
//听写结果回调接口(返回Json格式结果，用户可参见附录)；
//一般情况下会通过onResults接口多次返回结果，完整的识别内容是多次结果的累加；
//关于解析Json的代码可参见MscDemo中JsonParser类；
//isLast等于true时会话结束。
public void onResult(RecognizerResult results, boolean isLast){
DebugLog.Log("Result:"+results.getResultString ());
}
//会话发生错误回调接口
public void onError(SpeechError error) {
error.getPlainDescription(true) //获取错误码描述
}
//开始录音
public void onBeginOfSpeech() {}
//音量值0~30
public void onVolumeChanged(int volume){}
//结束录音
public void onEndOfSpeech() {}
//扩展用接口
public void onEvent(int eventType,int arg1,int arg2,String msg) {}
};
~~~
## 使用音频文件听写示例
~~~
//1.创建SpeechRecognizer对象
SpeechRecognizer mIat= SpeechRecognizer.createRecognizer( );
//2.设置听写参数，详见《MSC Reference Manual》SpeechConstant类
mIat.setParameter(SpeechConstant.DOMAIN, "iat");
mIat.setParameter(SpeechConstant.LANGUAGE, "zh_cn");
mIat.setParameter (SpeechConstant.ACCENT, "mandarin ");
mIat.setParameter(SpeechConstant.AUDIO_SOURCE, "-1");
//3.开始听写
mIat.startListening(mRecoListener);
//voiceBuffer为音频数据流，splitBuffer为自定义分割接口，将其以4.8k字节分割成数组
ArrayList<byte[]> buffers = splitBuffer(voiceBuffer,voiceBuffer.length, 4800);
for (int i = 0; i < buffers.size(); i++) {
// 每次写入msc数据4.8K,相当150ms录音数据
mIat.writeAudio(buffers.get(i), 0, buffers.get(i).length);
}
mIat.stopListening();
//听写监听器
private RecognizerListener mRecoListener = new RecognizerListener(){
public void onResult(RecognizerResult results, boolean isLast) {
DebugLog.Log("Result:"+results.getResultString ());
}
//会话发生错误回调接口
public void onError(SpeechError error) {
error.getPlainDescription(true) //获取错误码描述
}
//开始录音
public void onBeginOfSpeech() {}
//音量值0~30
public void onVolumeChange( int volume ) {}
}
~~~
## 上传用户词表
上传用户词表可以提高词表内词汇的识别率，也可以提高语义的效果，每个用户终端设备对应一个词表，用户词表的格式及构造方法详见[《MSC Reference Manual》](http://mscdoc.xfyun.cn/java/api)UserWords类。
~~~
private void uploadUserWords() {
SpeechRecognizer recognizer = SpeechRecognizer.getRecognizer();
UserWords userwords = new UserWords(USER_WORDS);
recognizer.setParameter( SpeechConstant.DATA_TYPE, "userword" );
recognizer.updateLexicon("userwords", userwords.toString(), lexiconListener);
}
/**
* 词表上传监听器
*/
LexiconListener lexiconListener = new LexiconListener() {
@Override
public void onLexiconUpdated(String lexiconId, SpeechError error) {
if (error == null)
DebugLog.Log("*************上传成功*************");
else
DebugLog.Log("*************" + error.getErrorCode()+ "*************");
}
};
~~~
