vue使用讯飞语音webapi

---

前言

项目要求实现语音转文字，决定使用讯飞语音的webapi，在网上找了很多方法，修修改改，经过了一些坑，终于实现了想要的效果

---

一、新建公用文件夹内放入以下两个文件（IatRecorder.js，transcode.worker.js）

IatRecorder.js

/* eslint-disable no-undef */
const APPID = '6d1b5e69'
const API_SECRET = ' '
const API_KEY = ' '
import store from '@/store'
import CryptoJS from 'crypto-js'
import Worker from './transcode.worker.js'
const transWorker = new Worker()
let startTime = ''
let endTime = ''

function getWebSocketUrl() {
  return new Promise((resolve, reject) => {
    // 请求地址根据语种不同变化
    let url = 'wss://iat-api.xfyun.cn/v2/iat'
    const host = 'iat-api.xfyun.cn'
    const apiKey = API_KEY
    const apiSecret = API_SECRET
    const date = new Date().toGMTString()
    const algorithm = 'hmac-sha256'
    const headers = 'host date request-line'
    const signatureOrigin = `host: ${host}\ndate: ${date}\nGET /v2/iat HTTP/1.1`
    const signatureSha = CryptoJS.HmacSHA256(signatureOrigin, apiSecret)
    const signature = CryptoJS.enc.Base64.stringify(signatureSha)
    const authorizationOrigin = `api_key="${apiKey}", algorithm="${algorithm}", headers="${headers}", signature="${signature}"`
    const authorization = btoa(authorizationOrigin)
    url = `${url}?authorization=${authorization}&date=${date}&host=${host}`
    resolve(url)
    reject()
  })
}
const IatRecorder = class {
  constructor({ language, accent, appId } = {}) {
    this.closeVoiceFlag = 0
    const self = this
    this.status = 'null'
    this.language = language || 'zh_cn'
    this.accent = accent || 'mandarin'
    this.appId = appId || APPID
    // 记录音频数据
    this.audioData = []
    // 记录听写结果
    this.resultText = ''
    // wpgs下的听写结果需要中间状态辅助记录
    this.resultTextTemp = ''
    transWorker.onmessage = function(event) {
      // console.log("构造方法中",self.audioData)
      self.audioData.push(...event.data)
    }
  }

  // 修改录音听写状态
  setStatus(status) {
    this.onWillStatusChange &&
      this.status !== status &&
      this.onWillStatusChange(this.status, status)
    this.status = status
  }
  setResultText({ resultText, resultTextTemp } = {}) {
    this.onTextChange && this.onTextChange(resultTextTemp || resultText || '')
    resultText !== undefined && (this.resultText = resultText)
    resultTextTemp !== undefined && (this.resultTextTemp = resultTextTemp)
  }
  // 修改听写参数
  setParams({ language, accent } = {}) {
    language && (this.language = language)
    accent && (this.accent = accent)
  }
  // 连接websocket
  connectWebSocket() {
    return getWebSocketUrl().then(url => {
      let iatWS
      if ('WebSocket' in window) {
        iatWS = new WebSocket(url)
      } else if ('MozWebSocket' in window) {
        iatWS = new MozWebSocket(url)
      } else {
        alert('浏览器不支持WebSocket')
        return
      }
      this.webSocket = iatWS
      this.setStatus('init')
      iatWS.onopen = e => {
        this.setStatus('ing')
        // 重新开始录音
        setTimeout(() => {
          this.webSocketSend()
        }, 500)
      }
      iatWS.onmessage = e => {
        this.result(e.data)
      }
      iatWS.onerror = e => {
        this.recorderStop()
      }
      iatWS.onclose = e => {
        endTime = Date.parse(new Date())
        // console.log('持续时间', endTime - startTime)
        this.recorderStop()
      }
    })
  }
  // 初始化浏览器录音
  recorderInit() {
    navigator.getUserMedia =
      navigator.getUserMedia ||
      navigator.webkitGetUserMedia ||
      navigator.mozGetUserMedia ||
      navigator.msGetUserMedia

    // 创建音频环境
    try {
      this.audioContext = new (window.AudioContext ||
        window.webkitAudioContext)()
      this.audioContext.resume()
      if (!this.audioContext) {
        alert('浏览器不支持webAudioApi相关接口')
        return
      }
    } catch (e) {
      if (!this.audioContext) {
        alert('浏览器不支持webAudioApi相关接口')
        return
      }
    }

    // 获取浏览器录音权限
    if (navigator.mediaDevices && navigator.mediaDevices.getUserMedia) {
      navigator.mediaDevices
        .getUserMedia({
          audio: true,
          video: false,
        })
        .then(stream => {
          getMediaSuccess(stream)
        })
        .catch(e => {
          getMediaFail(e)
        })
    } else if (navigator.getUserMedia) {
      navigator.getUserMedia(
        {
          audio: true,
          video: false,
        },
        stream => {
          getMediaSuccess(stream)
        },
        function(e) {
          getMediaFail1(e)
        },
      )
    } else {
      if (
        navigator.userAgent.toLowerCase().match(/chrome/) &&
        location.origin.indexOf('https://') < 0
      ) {
        alert(
          'chrome下获取浏览器录音功能，因为安全性问题，需要在localhost或127.0.0.1或https下才能获取权限',
        )
      } else {
        alert('无法获取浏览器录音功能，请升级软件版本')
      }
      this.voiceDialog = 0
      this.audioContext && this.audioContext.close()
      return
    }
    // 获取浏览器录音权限成功的回调
    const getMediaSuccess = stream => {
      this.voiceDialog = 1
      // console.log('getMediaSuccess')
      // 创建一个用于通过JavaScript直接处理音频
      this.scriptProcessor = this.audioContext.createScriptProcessor(0, 1, 1)
      this.scriptProcessor.onaudioprocess = e => {
        // 去处理音频数据
        if (this.status === 'ing') {
          //  console.log(transWorker)
          transWorker.postMessage(e.inputBuffer.getChannelData(0))
          //  this.audioData.push(e.inputBuffer.getChannelData(0))
        }
      }
      // 创建一个新的MediaStreamAudioSourceNode 对象，使来自MediaStream的音频可以被播放和操作
      this.mediaSource = this.audioContext.createMediaStreamSource(stream)
      // 连接
      this.mediaSource.connect(this.scriptProcessor)
      this.scriptProcessor.connect(this.audioContext.destination)
      this.connectWebSocket()
    }

    const getMediaFail = e => {
      alert(e)
      this.voiceDialog = 0
      // console.log(e)
      this.audioContext && this.audioContext.close()
      this.audioContext = undefined
      // 关闭websocket
      if (this.webSocket && this.webSocket.readyState === 1) {
        this.webSocket.close()
      }
    }
    const getMediaFail1 = e => {
      alert('请求麦克风失败,请添加权限！')
      this.voiceDialog = 0
      // console.log(e)
      this.audioContext && this.audioContext.close()
      this.audioContext = undefined
      // 关闭websocket
      if (this.webSocket && this.webSocket.readyState === 1) {
        this.webSocket.close()
      }
    }
  }
  recorderStart() {
    if (!this.audioContext) {
      this.recorderInit()
    } else {
      this.audioContext.resume()
      this.connectWebSocket()
    }
  }
  // 暂停录音
  recorderStop() {
    // safari下suspend后再次resume录音内容将是空白，设置safari下不做suspend
    if (
      !(
        /Safari/.test(navigator.userAgent) && !/Chrome/.test(navigator.userAgen)
      )
    ) {
      this.audioContext && this.audioContext.suspend()
    }
    this.setStatus('end')
  }
  // 处理音频数据
  // transAudioData(audioData) {
  //   audioData = transAudioData.transaction(audioData)
  //   this.audioData.push(...audioData)
  // }
  // 对处理后的音频数据进行base64编码，
  toBase64(buffer) {
    let binary = ''
    const bytes = new Uint8Array(buffer)
    const len = bytes.byteLength
    for (let i = 0; i < len; i++) {
      binary += String.fromCharCode(bytes[i])
    }
    return window.btoa(binary)
  }
  // 向webSocket发送数据
  webSocketSend() {
    if (this.webSocket.readyState !== 1) {
      return
    }
    let audioData = this.audioData.splice(0, 1280)
    const params = {
      common: {
        app_id: this.appId,
      },
      business: {
        language: this.language, //小语种可在控制台--语音听写（流式）--方言/语种处添加试用
        domain: 'iat',
        accent: this.accent, //中文方言可在控制台--语音听写（流式）--方言/语种处添加试用
        vad_eos: 5000,
        dwa: 'wpgs',
      },
      data: {
        status: 0,
        format: 'audio/L16;rate=16000',
        encoding: 'raw',
        audio: this.toBase64(audioData),
      },
    }
    // console.log('参数language：', this.language)
    // console.log('参数accent：', this.accent)
    this.webSocket.send(JSON.stringify(params))
    startTime = Date.parse(new Date())
    this.handlerInterval = setInterval(() => {
      // websocket未连接
      if (this.webSocket.readyState !== 1) {
        console.log('websocket未连接')
        this.closeVoiceFlag = 1
        this.audioData = []
        clearInterval(this.handlerInterval)
        return
      }
      if (this.audioData.length === 0) {
        //点击暂停录音
        // console.log('自动关闭', this.status)
        if (this.status === 'end') {
          this.webSocket.send(
            JSON.stringify({
              data: {
                status: 2,
                format: 'audio/L16;rate=16000',
                encoding: 'raw',
                audio: '',
              },
            }),
          )
          this.audioData = []
          clearInterval(this.handlerInterval)
        }
        return false
      }
      audioData = this.audioData.splice(0, 1280)
      // 中间帧
      this.webSocket.send(
        JSON.stringify({
          data: {
            status: 1,
            format: 'audio/L16;rate=16000',
            encoding: 'raw',
            audio: this.toBase64(audioData),
          },
        }),
      )
    }, 40)
  }
  result(resultData) {
    // 识别结束
    const jsonData = JSON.parse(resultData)
    if (jsonData.data && jsonData.data.result) {
      const data = jsonData.data.result
      let str = ''
      const resultStr = ''
      const ws = data.ws
      for (let i = 0; i < ws.length; i++) {
        str = str + ws[i].cw[0].w
      }
      // console.log('识别的结果为：', str)
      // 开启wpgs会有此字段(前提：在控制台开通动态修正功能)
      // 取值为 "apd"时表示该片结果是追加到前面的最终结果；取值为"rpl" 时表示替换前面的部分结果，替换范围为rg字段
      if (data.pgs) {
        if (data.pgs === 'apd') {
          // 将resultTextTemp同步给resultText
          this.setResultText({
            resultText: this.resultTextTemp,
          })
        }
        // 将结果存储在resultTextTemp中
        this.setResultText({
          resultTextTemp: this.resultText + str,
        })
      } else {
        this.setResultText({
          resultText: this.resultText + str,
        })
      }
    }
    if (jsonData.code === 0 && jsonData.data.status === 2) {
      this.webSocket.close()
    }
    if (jsonData.code !== 0) {
      this.webSocket.close()
      // console.log(`${jsonData.code}:${jsonData.message}`)
    }
  }
  start() {
    this.recorderStart()
    this.setResultText({ resultText: '', resultTextTemp: '' })
  }
  stop() {
    this.recorderStop()
  }
}

export default IatRecorder

transcode.worker.js

self.onmessage = function(e) {
  transAudioData.transcode(e.data)
}
const transAudioData = {
  transcode(audioData) {
    let output = transAudioData.to16kHz(audioData)
    output = transAudioData.to16BitPCM(output)
    output = Array.from(new Uint8Array(output.buffer))
    self.postMessage(output)
    // return output
  },
  to16kHz(audioData) {
    const data = new Float32Array(audioData)
    const fitCount = Math.round(data.length * (16000 / 44100))
    const newData = new Float32Array(fitCount)
    const springFactor = (data.length - 1) / (fitCount - 1)
    newData[0] = data[0]
    for (let i = 1; i < fitCount - 1; i++) {
      const tmp = i * springFactor
      const before = Math.floor(tmp).toFixed()
      const after = Math.ceil(tmp).toFixed()
      const atPoint = tmp - before
      newData[i] = data[before] + (data[after] - data[before]) * atPoint
    }
    newData[fitCount - 1] = data[data.length - 1]
    return newData
  },
  to16BitPCM(input) {
    const dataLength = input.length * (16 / 8)
    const dataBuffer = new ArrayBuffer(dataLength)
    const dataView = new DataView(dataBuffer)
    let offset = 0
    for (let i = 0; i < input.length; i++, offset += 2) {
      const s = Math.max(-1, Math.min(1, input[i]))
      dataView.setInt16(offset, s < 0 ? s * 0x8000 : s * 0x7fff, true)
    }
    return dataView
  },
}

二、在main.js中引入文件

import IatRecorder from '@/scripts/js/IatRecorder'
const iatRecorder = new IatRecorder('en_us', 'mandarin', '6d1b5e69') //小语种-中文方言-appId
Vue.prototype.iatRecorder = iatRecorder

三、在页面中使用

// 开始识别
handleVoiceStart() {
  const readyState = this.iatRecorder.webSocket
    ? this.iatRecorder.webSocket.readyState
    : 3
  if (readyState != 3) {
    Toast('上一段进程未结束，请稍后再使用该功能！')
  } else {
    this.myvoice = this.comments
    console.log(this.myvoice)
    this.voiceShow = true
    this.iatRecorder.start()
    const voice = setInterval(() => {
      if (this.iatRecorder.closeVoiceFlag == 1) {
        Toast('未检测到语音！')
        this.voiceShow = false
        this.iatRecorder.closeVoiceFlag = 0
        clearInterval(voice)
      }
      if (this.iatRecorder.voiceDialog == 0) {
        this.voiceShow = false
        clearInterval(voice)
      }
    }, 40)
  }
},
// 结束识别
handleVoiceEnd() {
  this.iatRecorder.stop()
  this.iatRecorder.onTextChange = text => {
    this.searchData = text
    this.comments = this.myvoice + this.searchData
  }
  this.voiceShow = false
},