node-microphone provides microphone input in Node.js by spawning platform tools:

• Linux: arecord
• macOS: rec from SoX
• Windows: sox

It returns the underlying PCM stream and also exposes info, error, and optional data events through an EventEmitter API.

The current 0.2.0 release also improves a few rough edges from 0.1.6:

• consistent default output format across platforms with fileType: 'raw'
• bundled TypeScript declarations
• constructor validation for invalid option values
• automatic cleanup when the recorder process closes
• test coverage for command generation and lifecycle behavior

Install the audio capture tool for your platform before using the package.

• Linux: ALSA tools, for example sudo apt-get install alsa-utils
• macOS: SoX, with the rec command available in PATH
• Windows: SoX, with sox available in PATH

npm install node-microphone

TypeScript declarations are bundled with the package. The package now targets Node.js 12+.

Compared with 0.1.6, version 0.2.0 adds and changes:

• TypeScript support through bundled index.d.ts
• a new fileType option so Linux, macOS, and Windows can produce the same output type by default
• a new spawnOptions option for passing supported child_process.spawn() settings
• a new close event emitted when the recording process exits
• safer constructor validation for malformed options such as invalid endian, non-boolean useDataEmitter, or unsupported spawnOptions.stdio
• internal process cleanup so a Microphone instance can be started again after the child exits
• refreshed package metadata, repository links, and automated tests

const Microphone = require('node-microphone');

const mic = new Microphone({
    rate: '16000',
    channels: '1',
    fileType: 'raw',
});

const micStream = mic.startRecording();

if (micStream) {
	micStream.pipe(myWritableStream);
}

mic.on('info', (info) => {
	console.log(info.toString());
});

mic.on('error', (error) => {
	console.error(error);
});

setTimeout(() => {
	mic.stopRecording();
}, 3000);

If useDataEmitter is enabled, audio chunks are also emitted through mic.on('data', chunk) in addition to being available from the returned stream. Invalid option values throw synchronously from the constructor with a TypeError.

Creates a new microphone instance.

Notes:

• On Linux, the package uses arecord -f <format> -t <fileType> -D <device>.
• On macOS, the package uses rec -t <fileType> -.
• On Windows, the package uses sox -t waveaudio <device> ... -t <fileType> -.
• On macOS, when device is provided, the package passes it through SoX's AUDIODEV environment variable.
• spawnOptions.stdio is intentionally rejected because the library depends on piped stdout and stderr.

Starts the recording process and returns the spawned process stdout stream. If recording is already active, it returns undefined.

Stops the spawned recording process and clears the internal process reference.

• info: stderr output from the recording command as a Buffer
• error: process or stream error
• close: emitted when the recording process closes
• data: stdout audio chunk as a Buffer when useDataEmitter is enabled

import Microphone = require('node-microphone');

const mic = new Microphone({
    rate: 16000,
    channels: 1,
    fileType: 'wav',
});

mic.on('close', (code, signal) => {
	console.log(code, signal);
});

npm test

The test suite covers Linux, macOS, and Windows command construction using mocked child processes.

Thanks to ashishbajaj99 and vincentsaluzzo for related microphone modules that helped shape this package.

MIT. See LICENSE.