实时语音识别转文字(调用腾讯云语音识别接口)

2,119 阅读7分钟

实现实时语音识别转文字功能可以采用腾讯云语音识别接口 目前腾讯云语音识别有实时、语音、一句话等方案,我们采用实时语音方案,可以在腾讯云官网申请免费体验套餐进行测试

1639375344924.jpg

在控制台内可以看到使用记录以及右下角的帮助文档 按照文档方法调用api即可

1.在左侧秘钥处获取appid以及相关秘钥,保存在data内方便使用

    data() {
	return {
		secretId:'',
                appId:,
		secretKey:'',
		url:'asr.cloud.tencent.com/asr/v2/',
		socket: null,
	};

2.按照文档要求拼装参数,参数含义可以文档查看,同时对SecretKey使用 HmacSha1 算法进行加密并做 base64 编码处理

opens(){
	const timestamp = parseInt(new Date().getTime() / 1000) - 1;
	const params = {
                secretid: this.secretId,
                timestamp: timestamp,
		expired: timestamp + 60 * 60,
                nonce: timestamp,
		engine_model_type: '16k_zh',
		voice_id: timestamp.toString(),
		voice_format: 8,
		convert_num_mode:1,
		};
         // 处理参数为url
	const url=this.getUrl(this.url,params)	
	this.connect(url)
	},
getUrl(url,params){
        let newurl=url+this.appId+'?'+Object.keys(params)
                    .sort(function(a, b) {
				return a.localeCompare(b);
                    })
                    .map((key)=>{
			return encodeURIComponent(key)+'='
                        +encodeURIComponent(params[key])
                    })
                    .join('&')
           // 使用 HmacSha1 算法进行加密并做 base64 编码处理
          let baseurl=this.baseUrl(newurl)
          newurl=newurl+'&'+'signature'+'='
          +encodeURIComponent(baseurl)
         return newurl
  },
 baseUrl(newurl){
	const hash = CryptoJS.HmacSHA1(newurl, this.secretKey);
	return uni.arrayBufferToBase64(this.toUint8Array(hash));
},
toUint8Array(wordArray){
	// Shortcuts
	const words = wordArray.words;
	const sigBytes = wordArray.sigBytes;
				
	// Convert
        const u8 = new Uint8Array(sigBytes);
	for (let i = 0; i < sigBytes; i++) {
		u8[i] = (words[i >>> 2] >>> (24 - (i % 4) * 8)) & 0xff;
		}
	return u8;
	},
参数名称必填类型描述
secretidString腾讯云注册账号的密钥 SecretId,可通过 API 密钥管理页面 获取。
timestampInteger当前 UNIX 时间戳,单位为秒。如果与当前时间相差过大,会引起签名过期错误。
expiredInteger签名的有效期截止时间 UNIX 时间戳,单位为秒。expired 必须大于 timestamp 且 expired - timestamp 小于90天。
nonceInteger随机正整数。用户需自行生成,最长10位。
engine_model_typeString引擎模型类型。 电话场景: • 8k_en:电话 8k 英语; • 8k_zh:电话 8k 中文普通话通用; • 8k_zh_finance:电话 8k 金融领域模型; 非电话场景: • 16k_zh:16k 中文普通话通用; • 16k_en:16k 英语; • 16k_ca:16k 粤语; • 16k_ko:16k 韩语; • 16k_zh-TW:16k 中文普通话繁体; • 16k_ja:16k 日语; • 16k_wuu-SH:16k 上海话方言; • 16k_zh_medical 医疗; • 16k_en_game 英文游戏; • 16k_zh_court 法庭; • 16k_en_edu 英文教育; • 16k_zh_edu 中文教育; • 16k_th 泰语。
voice_idString16位 String 串作为每个音频的唯一标识,用户自己生成。
voice_formatInteger语音编码方式,可选,默认值为4。1:pcm;4:speex(sp);6:silk;8:mp3;12:wav;14:m4a(每个分片须是一个完整的 m4a 音频);16:aac。
needvadInteger0:关闭 vad,1:开启 vad。 如果语音分片长度超过60秒,用户需开启 vad。
hotword_idString热词 id。用于调用对应的热词表,如果在调用语音识别服务时,不进行单独的热词 id 设置,自动生效默认热词;如果进行了单独的热词 id 设置,那么将生效单独设置的热词 id。
customization_idString自学习模型 id。用于调用对应的自学习模型,如果在调用语音识别服务时,不进行单独的自学习模型 id 设置,自动生效默认自学习模型;如果进行了单独的自学习模型 id 设置,那么将生效单独设置的自学习模型 id。
filter_dirtyInteger是否过滤脏词(目前支持中文普通话引擎)。默认为0。0:不过滤脏词;1:过滤脏词;2:将脏词替换为 * 。
filter_modalInteger是否过语气词(目前支持中文普通话引擎)。默认为0。0:不过滤语气词;1:部分过滤;2:严格过滤 。
filter_puncInteger是否过滤句末的句号(目前支持中文普通话引擎)。默认为0。0:不过滤句末的句号;1:过滤句末的句号。
convert_num_modeInteger是否进行阿拉伯数字智能转换(目前支持中文普通话引擎)。0:不转换,直接输出中文数字,1:根据场景智能转换为阿拉伯数字,3: 打开数学相关数字转换。默认值为1。
word_infoInt是否显示词级别时间戳。0:不显示;1:显示,不包含标点时间戳,2:显示,包含标点时间戳。支持引擎 8k_en、8k_zh、8k_zh_finance、16k_zh、16k_en、16k_ca、16k_zh-TW、16k_ja、16k_wuu-SH,默认为0。
vad_silence_timeInteger语音断句检测阈值,静音时长超过该阈值会被认为断句(多用在智能客服场景,需配合 needvad = 1 使用),取值范围:240-2000,单位 ms,此参数建议不要随意调整,可能会影响识别效果,目前仅支持 8k_zh、8k_zh_finance、16k_zh 引擎模型。
signatureString接口签名参数。

3.使用socket进行链接

connect(url){
	console.log(url)
	this.socket=uni.connectSocket({
	url: 'wss://'+url,
        success: data => {
            console.log('socket connect result ', data);
	},
        fail: e => {
            this.loading(JSON.stringify(e));
            }
	});
        this.socket.onOpen(function (res) {
            console.log('WebSocket连接已打开!');
	});
	this.socket.onError(function (res) {
		console.log('WebSocket连接打开失败,请检查!');
	});
	this.socket.onMessage((res)=>{
		console.log(res.data)
                res.data=JSON.parse(res.data)
		console.log('收到服务器内容:' +res.data);
		console.log(res.data)
	if(res.data.code===0 && res.data.result){
		console.log('识别成功',res.data.result)
	}else if(res.data.code == 0){
		this.startRecorder();
		}else{
		console.log('识别失败',res.data)
	}
	})
                this.socket.onClose( (res)=> {
                console.log('WebSocket 已关闭!');
                this.socket = null;
                this.stopRecorder();
		});
	},
   closes(){
    this.socket.close();
    },

4.打开录音机

startRecorder(){
	console.log('打开录音机')
	if (this.recorder == null) {
	const recorder = (this.recorder = uni.getRecorderManager());
	recorder.onFrameRecorded(({ isLastFrame, frameBuffer }) => {
		console.log(frameBuffer)
	if (this.socket) {
		this.socket.send({
			data: frameBuffer
		});
							
	if (isLastFrame) {
		this.socket.send({
                        data: JSON.stringify({
                                type: 'end'
                                    })
				});
                            }
			}
		});
	recorder.onError(({ errMsg }) => {
	console.log('recorder error', errMsg);
	if (errMsg != "operateRecorder:fail:audio is stop, don't stop record again") {
			this.loading('启动失败:' + errMsg);
			// this.stopConnect();
			}
		});
	recorder.onStart(() => {
			console.log('recorder start');
			//this.loading('正在识别');
			this.startTimer();
			});
	recorder.onStop(({ tempFilePath }) => {
				console.log('recorder stop', tempFilePath);
				});
				recorder.onPause(e => {
					console.log('recorder pause');
				});
                                }
                                this.recorder.start({
					duration: 60 * 1000,
					format: 'mp3',
					frameSize: 1.25,
					sampleRate: 16000,
					numberOfChannels: 1
				});
			},

5.全部demo代码

<template>
	<view>
				<button @click="opens">打开</button>
				<button @click="closes">关闭</button>
	</view>
</template>

<script>
	import CryptoJS from './cryptojs.js'
	export default {
		data() {
			return {
				secretId:'',
				appId:,
				secretKey:'',
				url:'asr.cloud.tencent.com/asr/v2/',
				socket: null,
				recorder:null,
				duration: 0,
				durationTimer: null,
			};
		},
		
		methods: {
			opens(){
				const timestamp = parseInt(new Date().getTime() / 1000) - 1;
				const params = {
					secretid: this.secretId,
					timestamp: timestamp,
					expired: timestamp + 60 * 60,
					nonce: timestamp,
					engine_model_type: '16k_zh',
					voice_id: timestamp.toString(),
					voice_format: 8,
					convert_num_mode:1,
				};
				const url=this.getUrl(this.url,params)	
				this.connect(url)
			},
			getUrl(url,params){
				let newurl=url+this.appId+'?'+Object.keys(params)
				.sort(function(a, b) {
						return a.localeCompare(b);
				})
				.map((key)=>{
					return encodeURIComponent(key)+'='+encodeURIComponent(params[key])
				})
				.join('&')
				let baseurl=this.baseUrl(newurl)
				newurl=newurl+'&'+'signature'+'='+encodeURIComponent(baseurl)
				return newurl
			},
			baseUrl(newurl){
				const hash = CryptoJS.HmacSHA1(newurl, this.secretKey);
				return uni.arrayBufferToBase64(this.toUint8Array(hash));
			},
			toUint8Array(wordArray){
				// Shortcuts
				const words = wordArray.words;
				const sigBytes = wordArray.sigBytes;
				
				// Convert
				const u8 = new Uint8Array(sigBytes);
				for (let i = 0; i < sigBytes; i++) {
					u8[i] = (words[i >>> 2] >>> (24 - (i % 4) * 8)) & 0xff;
				}
				return u8;
			},
			connect(url){
				console.log(url)
				this.socket=uni.connectSocket({
				    url: 'wss://'+url,
				   success: data => {
				   	console.log('socket connect result ', data);
				   },
				   fail: e => {
				   	this.loading(JSON.stringify(e));
				   }
				});
				this.socket.onOpen(function (res) {
				  console.log('WebSocket连接已打开!');
				});
				this.socket.onError(function (res) {
				  console.log('WebSocket连接打开失败,请检查!');
				});
				this.socket.onMessage((res)=>{
					console.log(res.data)
					res.data=JSON.parse(res.data)
				  console.log('收到服务器内容:' +res.data);
				  console.log(res.data)
				  if(res.data.code===0 && res.data.result){
					  console.log('识别成功',res.data.result)
				  }else if(res.data.code == 0){
					  this.startRecorder();
				  }else{
					  console.log('识别失败',res.data)
				  }
				})
				this.socket.onClose( (res)=> {
				  console.log('WebSocket 已关闭!');
				  this.socket = null;
				  this.stopRecorder();
				});
			},
			closes(){
				this.socket.close();
			},
			startRecorder(){
				console.log('打开录音机')
				if (this.recorder == null) {
					const recorder = (this.recorder = uni.getRecorderManager());
					recorder.onFrameRecorded(({ isLastFrame, frameBuffer }) => {
						console.log(frameBuffer)
						if (this.socket) {
							this.socket.send({
								data: frameBuffer
							});
							
							if (isLastFrame) {
								this.socket.send({
									data: JSON.stringify({
										type: 'end'
									})
								});
							}
						}
					});
					recorder.onError(({ errMsg }) => {
						console.log('recorder error', errMsg);
						if (errMsg != "operateRecorder:fail:audio is stop, don't stop record again") {
							this.loading('启动失败:' + errMsg);
							// this.stopConnect();
						}
					});
					recorder.onStart(() => {
						console.log('recorder start');
						//this.loading('正在识别');
						this.startTimer();
					});
					recorder.onStop(({ tempFilePath }) => {
						console.log('recorder stop', tempFilePath);
				
					});
					recorder.onPause(e => {
						console.log('recorder pause');
					});
				}
				
				this.recorder.start({
					duration: 60 * 1000,
					format: 'mp3',
					frameSize: 1.25,
					sampleRate: 16000,
					numberOfChannels: 1
				});
			},
			stopRecorder(){
				if (this.recorder) {
					this.recorder.stop()
				}
			},
			startTimer: function() {
				this.duration = 0;
				this.durationTimer = setInterval(() => {
					this.duration += 1;
				}, 1000);
			},
			stopTimer: function() {
				if (this.durationTimer) {
					clearInterval(this.durationTimer);
					this.durationTimer = null;
				}
			},
			
			open: function() {
				this.$refs.asr.show();
			},
					
		}
	};
</script>

<style lang="scss"></style>

CryptoJS代码

  
/*
 * [js-sha1]
 *
 * @version 0.6.0
 * @copyright H, J-C 2018-9-28
 * @license MIT
 */

var CryptoJS = function (g, l) {
	var e = {}, d = e.lib = {}, m = function () { }, k = d.Base = {
		extend: function (a) {
			m.prototype = this;
			var c = new m;
			a && c.mixIn(a);
			c.hasOwnProperty("init") || (c.init = function () {
			c.$super.init.apply(this, arguments)
			});
			c.init.prototype = c;
			c.$super = this;
			return c
		},
		create: function () {
			var a = this.extend();
			a.init.apply(a, arguments);
			return a
		},
		init: function () { },
		mixIn: function (a) {
			for (var c in a) a.hasOwnProperty(c) && (this[c] = a[c]);
			a.hasOwnProperty("toString") && (this.toString = a.toString)
		},
		clone: function () {
			return this.init.prototype.extend(this)
		}
	},
	p = d.WordArray = k.extend({
		init: function (a, c) {
			a = this.words = a || [];
			this.sigBytes = c != l ? c : 4 * a.length
		},
		toString: function (a) {
		  	return (a || n).stringify(this)
		},
		concat: function (a) {
			var c = this.words,
				q = a.words,
				f = this.sigBytes;
			a = a.sigBytes;
			this.clamp();
			if (f % 4)
				for (var b = 0; b < a; b++) c[f + b >>> 2] |= (q[b >>> 2] >>> 24 - 8 * (b % 4) & 255) << 24 - 8 * ((f + b) % 4);
			else if (65535 < q.length)
				for (b = 0; b < a; b += 4) c[f + b >>> 2] = q[b >>> 2];
			else c.push.apply(c, q);
			this.sigBytes += a;
			return this
		},
		clamp: function () {
			var a = this.words,
			c = this.sigBytes;
			a[c >>> 2] &= 4294967295 << 32 - 8 * (c % 4);
			a.length = g.ceil(c / 4)
		},
		clone: function () {
			var a = k.clone.call(this);
			a.words = this.words.slice(0);
			return a
		},
		random: function (a) {
			for (var c = [], b = 0; b < a; b += 4) c.push(4294967296 * g.random() | 0);
			return new p.init(c, a)
		}
	}),
	b = e.enc = {}, n = b.Hex = {
		stringify: function (a) {
			var c = a.words;
			a = a.sigBytes;
			for (var b = [], f = 0; f < a; f++) {
				var d = c[f >>> 2] >>> 24 - 8 * (f % 4) & 255;
				b.push((d >>> 4).toString(16));
				b.push((d & 15).toString(16))
			}
			return b.join("")
		},
		parse: function (a) {
			for (var c = a.length, b = [], f = 0; f < c; f += 2) b[f >>> 3] |= parseInt(a.substr(f, 2), 16) << 24 - 4 * (f % 8);
			return new p.init(b, c / 2)
		}
	}, j = b.Latin1 = {
		stringify: function (a) {
		  var c = a.words;
		  a = a.sigBytes;
		  for (var b = [], f = 0; f < a; f++) b.push(String.fromCharCode(c[f >>> 2] >>> 24 - 8 * (f % 4) & 255));
		  return b.join("")
		},
		parse: function (a) {
		  for (var c = a.length, b = [], f = 0; f < c; f++) b[f >>> 2] |= (a.charCodeAt(f) & 255) << 24 - 8 * (f % 4);
		  return new p.init(b, c)
		}
	}, h = b.Utf8 = {
		stringify: function (a) {
			try {
				return decodeURIComponent(escape(j.stringify(a)))
			} catch (c) {
				throw Error("Malformed UTF-8 data");
			}
		},
		parse: function (a) {
		  	return j.parse(unescape(encodeURIComponent(a)))
		}
	},
	r = d.BufferedBlockAlgorithm = k.extend({
		reset: function () {
			this._data = new p.init;
			this._nDataBytes = 0
		},
		_append: function (a) {
			"string" == typeof a && (a = h.parse(a));
			this._data.concat(a);
			this._nDataBytes += a.sigBytes
		},
		_process: function (a) {
			var c = this._data,
			b = c.words,
			f = c.sigBytes,
			d = this.blockSize,
			e = f / (4 * d),
			e = a ? g.ceil(e) : g.max((e | 0) - this._minBufferSize, 0);
			a = e * d;
			f = g.min(4 * a, f);
			if (a) {
				for (var k = 0; k < a; k += d) this._doProcessBlock(b, k);
				k = b.splice(0, a);
				c.sigBytes -= f
			}
			return new p.init(k, f)
		},
		clone: function () {
			var a = k.clone.call(this);
			a._data = this._data.clone();
			return a
		},
		_minBufferSize: 0
	});
	d.Hasher = r.extend({
		cfg: k.extend(),
		init: function (a) {
			this.cfg = this.cfg.extend(a);
			this.reset()
		},
		reset: function () {
			r.reset.call(this);
			this._doReset()
		},
		update: function (a) {
			this._append(a);
			this._process();
			return this
		},
		finalize: function (a) {
			a && this._append(a);
			return this._doFinalize()
		},
		blockSize: 16,
		_createHelper: function (a) {
			return function (b, d) {
			return (new a.init(d)).finalize(b)
			}
		},
		_createHmacHelper: function (a) {
			return function (b, d) {
			return (new s.HMAC.init(a, d)).finalize(b)
			}
		}
	});
	var s = e.algo = {};
	return e
}(Math);

(function () {
	var g = CryptoJS,
		l = g.lib,
		e = l.WordArray,
		d = l.Hasher,
		m = [],
		l = g.algo.SHA1 = d.extend({
			_doReset: function () {
				this._hash = new e.init([1732584193, 4023233417, 2562383102, 271733878, 3285377520])
			},
			_doProcessBlock: function (d, e) {
				for (var b = this._hash.words, n = b[0], j = b[1], h = b[2], g = b[3], l = b[4], a = 0; 80 > a; a++) {
					if (16 > a) m[a] = d[e + a] | 0;
					else {
					var c = m[a - 3] ^ m[a - 8] ^ m[a - 14] ^ m[a - 16];
					m[a] = c << 1 | c >>> 31
					}
					c = (n << 5 | n >>> 27) + l + m[a];
					c = 20 > a ? c + ((j & h | ~j & g) + 1518500249) : 40 > a ? c + ((j ^ h ^ g) + 1859775393) : 60 > a ? c + ((j & h | j & g | h & g) - 1894007588) : c + ((j ^ h ^ g) - 899497514);
					l = g;
					g = h;
					h = j << 30 | j >>> 2;
					j = n;
					n = c
				}
				b[0] = b[0] + n | 0;
				b[1] = b[1] + j | 0;
				b[2] = b[2] + h | 0;
				b[3] = b[3] + g | 0;
				b[4] = b[4] + l | 0
			},
			_doFinalize: function () {
				var d = this._data,
					e = d.words,
					b = 8 * this._nDataBytes,
					g = 8 * d.sigBytes;
				e[g >>> 5] |= 128 << 24 - g % 32;
				e[(g + 64 >>> 9 << 4) + 14] = Math.floor(b / 4294967296);
				e[(g + 64 >>> 9 << 4) + 15] = b;
				d.sigBytes = 4 * e.length;
				this._process();
				return this._hash
			},
			clone: function () {
				var e = d.clone.call(this);
				e._hash = this._hash.clone();
				return e
			}
	});
	g.SHA1 = d._createHelper(l);
	g.HmacSHA1 = d._createHmacHelper(l)
})();

(function () {
	var g = CryptoJS,
		l = g.enc.Utf8;
		g.algo.HMAC = g.lib.Base.extend({
		init: function (e, d) {
			e = this._hasher = new e.init;
			"string" == typeof d && (d = l.parse(d));
			var g = e.blockSize,
			k = 4 * g;
			d.sigBytes > k && (d = e.finalize(d));
			d.clamp();
			for (var p = this._oKey = d.clone(), b = this._iKey = d.clone(), n = p.words, j = b.words, h = 0; h < g; h++) n[h] ^= 1549556828, j[h] ^= 909522486;
			p.sigBytes = b.sigBytes = k;
			this.reset()
		},
		reset: function () {
			var e = this._hasher;
			e.reset();
			e.update(this._iKey)
		},
		update: function (e) {
			this._hasher.update(e);
			return this
		},
		finalize: function (e) {
			var d = this._hasher;
			e = d.finalize(e);
			d.reset();
			return d.finalize(this._oKey.clone().concat(e))
		}
	})
})();
  
  //使用算法
  // var key = "f7205fffe445421fdssdfsdfdsfs"
  // var sha1_result = CryptoJS.HmacSHA1("helloword", key)
  // console.log('-------',sha1_result.toString())
  

export default CryptoJS;