audiotee/README.md

AudioTee

AudioTee captures your Mac's system audio output and writes PCM encoded chunks of it to stdout at regular intervals, either in base64-encoded JSON (good for humans, easy on terminals) or binary (good for other programs). It uses the Core Audio taps API introduced in macOS 14.2 (released in December 2023). You can do whatever you want with this audio - stream it somewhere else, save it to disk, visualize it, etc.

By default, it taps the audio output from all running process and selects the most appropriate audio chunk output format to use based on the presence of a tty. Tap output is forced to mono (not configurable) and preserves your output device's sample rate unless you pass a --sample-rate flag. Only the default output device is currently supported.

My original (and so far only) use case is streaming audio to a parent process which communicates with a realtime ASR service, so AudioTee makes some design decisions you might not agree with. Open an issue or a PR and we can talk about them. I'm also no Swift developer, so contributions improving codebase idioms and general hygiene are welcome.

Recording system audio is harder than it should be on macOS, and folks often wrestle with outdated advice and poorly documented APIs. It's a boring problem which stands in the way of lots of fun applications. There's more code here than you need to solve this problem yourself: the main classes of interest are probably Core/AudioTapManager and Core/AudioRecorder. Everything's wired together in CLI/AudioTee. The rest is just CLI configuration support, output formatting logic, and some utility functions you could probably live without.

Requirements

• macOS 14.2 or later
• Swift 5.9 or later (no need for XCode)
• System audio recording permissions (see below)

Quick start

git clone git@github.com:makeusabrew/audiotee.git
cd audiotee
swift run

Build

# omit '-c release' to get a debug build
swift build -c release

Usage

Basic usage

Replace the path below with .build/<arch>/<target>/audiotee, e.g. build/arm64-apple-macosx/release/audiotee for a release build on Apple Silicon.

# Auto-detect output format (JSON in terminal, binary when piped)
./audiotee

# Always use JSON format (terminal-safe)
./audiotee --format json

# Always use binary format (pipe-optimised)
./audiotee --format binary

Audio conversion

# Convert to 16kHz mono (useful for ASR services)
./audiotee --sample-rate 16000

# Other supported sample rates: 22050, 24000, 32000, 44100, 48000
./audiotee --sample-rate 44100

Tap configuration

For now, only a subset of the CATapDescription (https://developer.apple.com/documentation/coreaudio/capturing-system-audio-with-core-audio-taps) interface is exposed. PRs welcome.

# Tap all system audio (default)
./audiotee

# Tap only a specific process (by PID)
./audiotee --include-processes 1234

# Tap multiple specific processes
./audiotee --include-processes 1234 5678 9012

# Tap everything *except* a specific process (by PID)
./audiotee --exclude-processes 1234

# Exclude multiple specific processes
./audiotee --exclude-processes 1234 5678 9012

# Mute processes being tapped (so they don't play through speakers)
./audiotee --mute

# Custom chunk duration (default 0.2 seconds, max 5.0)
./audiotee --chunk-duration 0.1

Output formats

AudioTee supports two output formats optimised for different use cases:

JSON format (--format json or auto in terminal)

JSON messages to stdout, one per line. Audio data is base64-encoded for terminal safety.

Binary format (--format binary or auto when piped)

JSON metadata lines followed by raw binary audio data. More efficient for piping to other processes.

Protocol

Message types

All messages (except raw binary audio chunks) follow this envelope structure:

{
  "timestamp": "2024-03-21T15:30:45.123Z",
  "message_type": "...",
  "data": { ... }
}

1. Metadata

Sent once at startup to describe the audio format:

{
  "timestamp": "2024-03-21T15:30:45.123Z",
  "message_type": "metadata",
  "data": {
    "sample_rate": 48000,
    "channels_per_frame": 1,
    "bits_per_channel": 32,
    "is_float": true,
    "capture_mode": "audio",
    "device_name": null,
    "device_uid": null,
    "encoding": "pcm_f32le"
  }
}

2. Stream start

Indicates audio data will follow:

{
  "timestamp": "2024-03-21T15:30:45.123Z",
  "message_type": "stream_start",
  "data": null
}

3. Audio data

JSON format:

{
  "timestamp": "2024-03-21T15:30:45.123Z",
  "message_type": "audio",
  "data": {
    "timestamp": "2024-03-21T15:30:45.123Z",
    "duration": 0.2,
    "peak_amplitude": 0.45,
    "audio_data": "base64_encoded_raw_audio..."
  }
}

Binary format:

{
  "timestamp": "2024-03-21T15:30:45.123Z",
  "message_type": "audio",
  "data": {
    "timestamp": "2024-03-21T15:30:45.123Z",
    "duration": 0.2,
    "peak_amplitude": 0.45,
    "audio_length": 9600
  }
}

Followed immediately by 9600 bytes of raw binary audio data

4. Stream stop

Sent when recording stops:

{
  "timestamp": "2024-03-21T15:30:45.123Z",
  "message_type": "stream_stop",
  "data": null
}

5. Log messages

Info, error, and debug messages (useful for monitoring):

{
  "timestamp": "2024-03-21T15:30:45.123Z",
  "message_type": "info",
  "data": {
    "message": "Starting AudioTee...",
    "context": { "output_format": "auto" }
  }
}

Consuming output

JSON format:

1. Parse each line as JSON using the envelope structure
2. Use metadata message to understand the audio format
3. For audio messages, decode audio_data from base64 to get raw PCM data
4. Do something with each chunk of data

Binary format:

1. Parse JSON metadata lines using the envelope structure
2. Use metadata message to understand the audio format
3. For audio messages, read audio_length bytes of raw binary data after the JSON line
4. Do something with each chunk of data

Note: binary is actually a mixed mode; JSON during boot, JSON packet header information preceding each binary chunk.

Command Line options

• --format, -f: Output format (json, binary, auto) [default: auto]
• --include-processes: Process IDs to tap (space-separated, empty = all processes)
• --exclude-processes: Process IDs to exclude (space-separated, empty = none)
• --mute: Mute processes being tapped
• --sample-rate: Target sample rate (8000, 16000, 22050, 24000, 32000, 44100, 48000)
• --chunk-duration: Audio chunk duration in seconds [default: 0.2, max: 5.0]

Permissions

There is no provision in the code to pre-emptively check for the required NSAudioCaptureUsageDescription permission, so you'll be prompted the first time AudioTee tries to record anything. If you want to check and/or request permissions ahead of time, check out AudioCap's clever TCC probing approach.

References

• Apple Core Audio Taps Documentation
• AudioCap Implementation

License

The MIT License

Copyright (C) 2025 Nick Payne.