以前的问题已经提出了这个相同或相似的问题

• Can Web Speech API used in conjunction with Web Audio API?

• How to access audio result from Speech Synthesis API?

• Record HTML5 SpeechSynthesisUtterance generated speech to file

• generate audio file with W3C Web Speech API

但似乎没有使用 window.speechSynthesis() 创建解决方法。尽管有使用 epeak 的解决方法，meSpeak How to create or convert text to audio at chromium browser?或向外部服务器发出请求。

如何捕获和记录 window.speechSynthesis.speak() 调用的音频输出并将结果作为 Blob、ArrayBuffer、AudioBuffer 还是其他对象类型？

最佳答案

Web Speech API Specification目前不提供有关如何实现返回或捕获和录制 window.speechSynthesis.speak() 调用的音频输出的方法或提示。

另见

• MediaStream, ArrayBuffer, Blob audio result from speak() for recording?

• Re: MediaStream, ArrayBuffer, Blob audio result from speak() for recording?

• Re: MediaStream, ArrayBuffer, Blob audio result from speak() for recording? .在相关部分，用例包括但不限于

  1. 说话有问题的人；例如，遭受过 中风或其他抑制疾病的交流。他们可以转换 文本到音频文件并将文件发送给另一个人或组。 此功能将有助于他们与其他人交流， 类似于帮助史蒂芬霍金交流的技术；

  2. 目前，唯一能听到音频输出的人是 在浏览器前；本质上，没有充分利用 文本到语音功能。音频结果可以用作 电子邮件中的附件；媒体流；聊天系统；或其他 通讯应用。即控制生成的音频输出；

  3. 另一个应用是提供免费、自由、开源的音频 字典和翻译服务——客户端到客户端和客户端到服务器， 服务器到客户端。

可以使用 navigator.mediaDevices.getUserMedia() 和 MediaRecorder( )。 Chromium 浏览器返回预期结果。 Firefox 的实现有 issues .在 navigator.mediaDevices.getUserMedia() 提示符下选择 Monitor of Built-in Audio Analog Stereo。

解决方法很麻烦。我们应该能够获得生成的音频，至少作为 Blob，无需 navigator.mediaDevices.getUserMedia() 和 MediaRecorder()。

浏览器用户、JavaScript 和 C++ 开发人员、浏览器实现者和规范作者显然需要更多的兴趣来进一步投入；为该功能创建适当的规范，并在浏览器的源代码中实现一致；见How to implement option to return Blob, ArrayBuffer, or AudioBuffer from window.speechSynthesis.speak() call .

在 Chromium 中，应安装一个语音调度程序并使用 --enable-speech-dispatcher 标志集启动实例，因为 window.speechSynthesis.getVoices() 返回空数组，参见 How to use Web Speech API at chromium? .

概念验证

// SpeechSynthesisRecorder.js guest271314 6-17-2017
// Motivation: Get audio output from `window.speechSynthesis.speak()` call
// as `ArrayBuffer`, `AudioBuffer`, `Blob`, `MediaSource`, `MediaStream`, `ReadableStream`, or other object or data types
// See https://lists.w3.org/Archives/Public/public-speech-api/2017Jun/0000.html
// https://github.com/guest271314/SpeechSynthesisRecorder

// Configuration: Analog Stereo Duplex
// Input Devices: Monitor of Built-in Audio Analog Stereo, Built-in Audio Analog Stereo

class SpeechSynthesisRecorder {
  constructor({text = "", utteranceOptions = {}, recorderOptions = {}, dataType = ""}) {
    if (text === "") throw new Error("no words to synthesize");
    this.dataType = dataType;
    this.text = text;
    this.mimeType = MediaRecorder.isTypeSupported("audio/webm; codecs=opus") 
                    ? "audio/webm; codecs=opus" : "audio/ogg; codecs=opus";
    this.utterance = new SpeechSynthesisUtterance(this.text);
    this.speechSynthesis = window.speechSynthesis;
    this.mediaStream_ = new MediaStream();
    this.mediaSource_ = new MediaSource();
    this.mediaRecorder = new MediaRecorder(this.mediaStream_, {
      mimeType: this.mimeType,
      bitsPerSecond: 256 * 8 * 1024
    });
    this.audioContext = new AudioContext();
    this.audioNode = new Audio();
    this.chunks = Array();
    if (utteranceOptions) {
      if (utteranceOptions.voice) {
        this.speechSynthesis.onvoiceschanged = e => {
          const voice = this.speechSynthesis.getVoices().find(({
            name: _name
          }) => _name === utteranceOptions.voice);
          this.utterance.voice = voice;
          console.log(voice, this.utterance);
        }
        this.speechSynthesis.getVoices();
      }
      let {
        lang, rate, pitch
      } = utteranceOptions;
      Object.assign(this.utterance, {
        lang, rate, pitch
      });
    }
    this.audioNode.controls = "controls";
    document.body.appendChild(this.audioNode);
  }
  start(text = "") {
    if (text) this.text = text;
    if (this.text === "") throw new Error("no words to synthesize");
    return navigator.mediaDevices.getUserMedia({
        audio: true
      })
      .then(stream => new Promise(resolve => {
        const track = stream.getAudioTracks()[0];
        this.mediaStream_.addTrack(track);
        // return the current `MediaStream`
        if (this.dataType && this.dataType === "mediaStream") {
          resolve({tts:this, data:this.mediaStream_});
        };
        this.mediaRecorder.ondataavailable = event => {
          if (event.data.size > 0) {
            this.chunks.push(event.data);
          };
        };
        this.mediaRecorder.onstop = () => {
          track.stop();
          this.mediaStream_.getAudioTracks()[0].stop();
          this.mediaStream_.removeTrack(track);
          console.log(`Completed recording ${this.utterance.text}`, this.chunks);
          resolve(this);
        }
        this.mediaRecorder.start();
        this.utterance.onstart = () => {
          console.log(`Starting recording SpeechSynthesisUtterance ${this.utterance.text}`);
        }
        this.utterance.onend = () => {
          this.mediaRecorder.stop();
          console.log(`Ending recording SpeechSynthesisUtterance ${this.utterance.text}`);
        }
        this.speechSynthesis.speak(this.utterance);
      }));
  }
  blob() {
    if (!this.chunks.length) throw new Error("no data to return");
    return Promise.resolve({
      tts: this,
      data: this.chunks.length === 1 ? this.chunks[0] : new Blob(this.chunks, {
        type: this.mimeType
      })
    });
  }
  arrayBuffer(blob) {
    if (!this.chunks.length) throw new Error("no data to return");
    return new Promise(resolve => {
      const reader = new FileReader;
      reader.onload = e => resolve(({
        tts: this,
        data: reader.result
      }));
      reader.readAsArrayBuffer(blob ? new Blob(blob, {
        type: blob.type
      }) : this.chunks.length === 1 ? this.chunks[0] : new Blob(this.chunks, {
        type: this.mimeType
      }));
    });
  }
  audioBuffer() {
    if (!this.chunks.length) throw new Error("no data to return");
    return this.arrayBuffer()
      .then(ab => this.audioContext.decodeAudioData(ab))
      .then(buffer => ({
        tts: this,
        data: buffer
      }))
  }
  mediaSource() {
    if (!this.chunks.length) throw new Error("no data to return");
    return this.arrayBuffer()
      .then(({
        data: ab
      }) => new Promise((resolve, reject) => {
        this.mediaSource_.onsourceended = () => resolve({
          tts: this,
          data: this.mediaSource_
        });
        this.mediaSource_.onsourceopen = () => {
          if (MediaSource.isTypeSupported(this.mimeType)) {
            const sourceBuffer = this.mediaSource_.addSourceBuffer(this.mimeType);
            sourceBuffer.mode = "sequence"
            sourceBuffer.onupdateend = () =>
              this.mediaSource_.endOfStream();
            sourceBuffer.appendBuffer(ab);
          } else {
            reject(`${this.mimeType} is not supported`)
          }
        }
        this.audioNode.src = URL.createObjectURL(this.mediaSource_);
      }));
  }
  readableStream({size = 1024, controllerOptions = {}, rsOptions = {}}) {
    if (!this.chunks.length) throw new Error("no data to return");
    const src = this.chunks.slice(0);
    const chunk = size;
    return Promise.resolve({
      tts: this,
      data: new ReadableStream(controllerOptions || {
        start(controller) {
            console.log(src.length);
            controller.enqueue(src.splice(0, chunk))
          },
          pull(controller) {
            if (src.length = 0) controller.close();
            controller.enqueue(src.splice(0, chunk));
          }
      }, rsOptions)
    });
  }
}

用法

let ttsRecorder = new SpeechSynthesisRecorder({
   text: "The revolution will not be televised", 
   utternanceOptions: {
     voice: "english-us espeak",
     lang: "en-US",
     pitch: .75,
     rate: 1
   }
 });

 // ArrayBuffer
 ttsRecorder.start()
 // `tts` : `SpeechSynthesisRecorder` instance, `data` : audio as `dataType` or method call result
 .then(tts => tts.arrayBuffer())
 .then(({tts, data}) => {
   // do stuff with `ArrayBuffer`, `AudioBuffer`, `Blob`,
   // `MediaSource`, `MediaStream`, `ReadableStream`
   // `data` : `ArrayBuffer`
   tts.audioNode.src = URL.createObjectURL(new Blob([data], {type:tts.mimeType}));
   tts.audioNode.title = tts.utterance.text;
   tts.audioNode.onloadedmetadata = () => {
     console.log(tts.audioNode.duration);
     tts.audioNode.play();
   }
 })
 // AudioBuffer     
 ttsRecorder.start()
 .then(tts => tts.audioBuffer())
 .then(({tts, data}) => {
   // `data` : `AudioBuffer`
   let source = tts.audioContext.createBufferSource();
   source.buffer = data;
   source.connect(tts.audioContext.destination);
   source.start()
 })
 // Blob
 ttsRecorder.start()
 .then(tts => tts.blob())
 .then(({tts, data}) => {
   // `data` : `Blob`
   tts.audioNode.src = URL.createObjectURL(blob);
   tts.audioNode.title = tts.utterance.text;
   tts.audioNode.onloadedmetadata = () => {
     console.log(tts.audioNode.duration);
     tts.audioNode.play();
   }
 })
 // ReadableStream
 ttsRecorder.start()
 .then(tts => tts.readableStream())
 .then(({tts, data}) => {
   // `data` : `ReadableStream`
   console.log(tts, data);
   data.getReader().read().then(({value, done}) => {
     tts.audioNode.src = URL.createObjectURL(value[0]);
     tts.audioNode.title = tts.utterance.text;
     tts.audioNode.onloadedmetadata = () => {
       console.log(tts.audioNode.duration);
       tts.audioNode.play();
     }
   })
 })
 // MediaSource
 ttsRecorder.start()
 .then(tts => tts.mediaSource())
 .then(({tts, data}) => {
   console.log(tts, data);
   // `data` : `MediaSource`
   tts.audioNode.srcObj = data;
   tts.audioNode.title = tts.utterance.text;
   tts.audioNode.onloadedmetadata = () => {
     console.log(tts.audioNode.duration);
     tts.audioNode.play();
   }
 })
 // MediaStream
 let ttsRecorder = new SpeechSynthesisRecorder({
   text: "The revolution will not be televised", 
   utternanceOptions: {
     voice: "english-us espeak",
     lang: "en-US",
     pitch: .75,
     rate: 1
   }, 
   dataType:"mediaStream"
 });
 ttsRecorder.start()
 .then(({tts, data}) => {
   // `data` : `MediaStream`
   // do stuff with active `MediaStream`
 })
 .catch(err => console.log(err))

plnkr

关于javascript - 如何从 window.speechSynthesis.speak() 调用中捕获生成的音频？，我们在Stack Overflow上找到一个类似的问题： https://stackoverflow.com/questions/45003548/