在web端，基于JavaScript用olami开放平台sdk也可以实现语音识别语义理解。本文就实现了这样一个小程序，web客户端本地用麦克风录音，录音的数据用speex压缩，然后跨域向服务器发送请求，返回识别的语音和语义字符串并显示。 先上图： 如下图刚载入的时候，未录音前界面
点击开始录音button后 一句话说完自动检测尾音结束标志然后压缩上传给服务器进行识别 将从服务器获取的识别结果显示到界面上
本例中说的语音是：“我要听三国演义这本书”，用的是android平台听书app建立的语法。
返回的json字串如下： { “data”: { “asr”: { “result”: “我要听三国演义这本书”, “speech_status”: 0, “final”: true, “status”: 0 }, “nli”: [ { “desc_obj”: { “result”: “正在努力搜索中，请稍等”, “status”: 0 }, “semantic”: [ { “app”: “musiccontrol”, “input”: “我要听三国演义这本书”, “slots”: [ { “name”: “songname”, “value”: “三国演义” } ], “modifier”: [ “play” ], “customer”: “58df512384ae11f0bb7b487e” } ], “type”: “musiccontrol” } ] }, “status”: “ok” } 下面是建立的工程目录结构，发布后，网页打开运行在chrome或者QQ浏览器均可。
下面讲述下voiceRecognize.html这个文件，其他都是min.js,只需知道如何调用就可以了。 <!DOCTYPE HTML> <html> <head> <meta charset="utf-8"> <title>voice recognize test</title> <script type="text/javascript" src="jquery-1.10.1.min.js"></script> <script type="text/javascript" src="jQuery.md5.js"></script> <script src="speex.min.js"></script> <script src="pcmdata.min.js"></script> </head> <body onload= "load()"> <fieldset style="width:50%;margin:10px auto"> <p align="center" >点击开始button录音,点击停止button停止录音并进行识别</p> <p align="center"> <input type="button" id="recordbutton" value="开始录音"onclick="StartRecording()"/> <input type="button" id="speexEncodebutton" value="停止录音"onclick="speexEncode()"/> </p> <p id ="result" style="margin-bottom:100px;margin-top:50px">result:</p> </p> <script src="recorder.min.js"></script> <script src="uploaddata.min.js"></script> <script> window.AudioContext = window.AudioContext || window.webkitAudioContext; var audioContext = new AudioContext(); var audioInput = null, realAudioInput = null, inputPoint = null, audioRecorder = null; var rafID = null; var analyserContext = null; var recIndex = 0; var recording = false; var bRecorded = false; function load(){ initAudio();//初始化recorder setAuthorization("http://cn.olami.ai/cloudservice/api","51a4bb56ba954655a4fc834bfdc46af1","asr","68bff251789b426896e70e888f919a6d","nli"); setCallBackFromServerResult(getResultFromServer); } function initAudio() { if (!navigator.getUserMedia) navigator.getUserMedia = navigator.webkitGetUserMedia || navigator.mozGetUserMedia; if (!navigator.cancelAnimationFrame) navigator.cancelAnimationFrame = navigator.webkitCancelAnimationFrame || navigator.mozCancelAnimationFrame; if (!navigator.requestAnimationFrame) navigator.requestAnimationFrame = navigator.webkitRequestAnimationFrame || navigator.mozRequestAnimationFrame; navigator.getUserMedia({audio:true}, gotStream, function(e) { alert('Error getting audio'); console.log(e); }); } function gotStream(stream) { inputPoint = audioContext.createGain(); // Create an AudioNode from the stream. realAudioInput = audioContext.createMediaStreamSource(stream); audioInput = realAudioInput; audioInput.connect(inputPoint); audioRecorder = new Recorder( inputPoint ); } function StartRecording() { // start recording if (audioRecorder == null) { initAudio(); alert("need initialize media"); } audioRecorder.clear(); audioRecorder.record(); recording = true; bRecorded = false; ToggleLabels(); RegisterCallBackToRecorder(); } function StopRecording() { audioRecorder.stop(); audioRecorder.getBuffers( gotBuffers ); } function RegisterCallBackToRecorder() {//检测语音结束后回调 audioRecorder.setCallBack(speexEncode); } function ToggleLabels() { if(recording) { document .getElementById("recordbutton").value = "录音中ֹ"; document .getElementById("speexEncodebutton").value = "停止录音"; var btn = document .getElementById("recordbutton").value; }else{ document .getElementById("speexEncodebutton").value = "识别中"; document .getElementById("recordbutton").value = "停止录音ֹ"; } } window.record = function(e) { if(!recording) { StartRecording(); recording = true; bRecorded = false; } else { StopRecording(); recording = false; bRecorded = true; } ToggleLabels(); }; window.speexEncode = function() { exportSpeex(); }; function exportSpeex() { recording = false; bRecorded = true; ToggleLabels(); audioRecorder.stop(); audioRecorder.exportPCM(uploadSpeexData); } function getResultFromServer() { document .getElementById('result').innerText = JSON.stringify(result); document .getElementById("speexEncodebutton").value = "停止录音"; document .getElementById("recordbutton").value = "开始录音ֹ" } </script> </body> </html> 浏览器载入的时候，先调用load()进行初始化 function load(){ setAuthorization( "http://cn.olami.ai/cloudservice/api",//serverurl "51a4bb56ba954655a4fc834bfdc46af1", //appkey "asr",//api类型 "68bff251789b426896e70e888f919a6d",//appSecret "nli");//seq setCallBackFromServerResult(getResultFromServer); } initAudio()中初始化了recorder用于获取麦克风资源做录音使用。 setAuthorization函数，参数分别是 url 服务器地址
appkey 开放平台注册应用后获得的appkey
api api类型选asr为语音
appSecret 开放平台注册应用后获得的appSecret
nli 为seq表示包含语音语义返回，为stt表示只有语音返回 setCallBackFromServerResult(getResultFromServer) 注册录音介绍并且识别出结果后的回调，在回调函数中可以把结果输出到界面上。 当点击开始录音button后，调用 function StartRecording() { // start recording if (audioRecorder == null) { initAudio(); alert("need initialize media"); } audioRecorder.clear(); audioRecorder.record(); recording = true; bRecorded = false; ToggleLabels(); RegisterCallBackToRecorder(); //注册recoder回调 } 当在录音的js代码中，会自动检测尾音结束，然后回调注册的函数speexEncode ()，点击停止录音button一样调用这个函数 window.speexEncode = function() { exportSpeex(); }; function exportSpeex() { recording = false; bRecorded = true; ToggleLabels();//更新界面的button状态 audioRecorder.stop(); audioRecorder.exportPCM(uploadSpeexData); } audioRecorder.exportPCM(uploadSpeexData) 实现了将录好的16Kpcm语音压缩成speex格式并上传到服务器，从服务器取得结果后调用注册的setCallBackFromServerResult(getResultFromServer)函数，然后再函数getResultFromServer中进行结果的输出显示。 代码下载地址： 相关网站链接：
| 
收藏
淘帖
踩
奈斯
