If you just want to get some audio data from your microphone, this is what you're looking for!

Converts a MediaStream (from getUserMedia) into a standard Node.js-style stream for easy pipe()ing.

Note: This only works in a limited set of browsers (typically with webpack or browserify), and then only for https or localhost in Chrome. It does not work in Node.js.

const getUserMedia = require('get-user-media-promise');
const MicrophoneStream = require('microphone-stream').default;

document.getElementById('my-start-button').onclick = function() {

  // Note: in most browsers, this constructor must be called in response to a click/tap, 
  // or else the AudioContext will remain suspended and will not provide any audio data.
  const micStream = new MicrophoneStream();

  getUserMedia({ video: false, audio: true })
    .then(function(stream) {
      micStream.setStream(stream);
    }).catch(function(error) {
      console.log(error);
    });

  // get Buffers (Essentially a Uint8Array DataView of the same Float32 values)
  micStream.on('data', function(chunk) {
    // Optionally convert the Buffer back into a Float32Array
    // (This actually just creates a new DataView - the underlying audio data is not copied or modified.)
    const raw = MicrophoneStream.toRaw(chunk)
    //...

    // note: if you set options.objectMode=true, the `data` event will output AudioBuffers instead of Buffers
   });

  // or pipe it to another stream
  micStream.pipe(/*...*/);

  // Access the internal audioInput for connecting to another nodes
  micStream.audioInput.connect(/*...*/));

  // It also emits a format event with various details (frequency, channels, etc)
  micStream.on('format', function(format) {
    console.log(format);
  });

  // Stop when ready
  document.getElementById('my-stop-button').onclick = function() {
    micStream.stop();
  };
 }

Where opts is an option object, with defaults:

{
  stream: null,
  objectMode: false,
  bufferSize: null,
  context: null
}

• stream: MediaStream instance. For iOS compatibility, it is recommended that you create the MicrophoneStream instance in response to the tap - before you have a MediaStream, and then later call setStream() with the MediaStream.

• bufferSize: Possible values: null, 256, 512, 1024, 2048, 4096, 8192, 16384. From Mozilla's Docs:

  It is recommended for authors to not specify this buffer size and allow the implementation to pick a good buffer size to balance between latency and audio quality.

• objectMode: if true, stream enters objectMode and emits AudioBuffers instead of Buffers. This has implications for pipe()'ing to other streams.

• context is the AudioContext instance. If omitted, one will be created automatically.

Set the mediaStream, necessary for iOS 11 support where the underlying AudioContext must be resumed in response to a user tap, but the mediaStream won't be available until later. Note: Some versions of Firefox leave the recording icon in place after recording has stopped.

Stops the recording. Note: Some versions of Firefox leave the recording icon in place after recording has stopped.

Temporarily stop emitting new data. Audio data recieved from the microphone after this will be dropped.

Note: the underlying Stream interface has a .pause() API that causes new data to be buffered rather than dropped.

Resume emitting new audio data after pauseRecording() was called.

Emits either a Buffer with raw 32-bit Floating point audio data, or if objectMode is set, an AudioBuffer containing the data + some metadata.

One-time event with details of the audio format. Example:

{
  channels: 1,
  bitDepth: 32,
  sampleRate: 48000,
  signed: true,
  float: true
}

Note: Here is an example of converting WebAudio data to L16 format. (This is the format most commonly used for the data portion of .wav files; see RFC 1890 section 4.4.8).

Converts a Buffer (from a data event or from calling .read()) back to the original Float32Array DataView format. (The underlying audio data is not copied or modified.)