@acpfx/mic-speaker

Native microphone capture with acoustic echo cancellation (AEC). Uses OS-level audio APIs via Rust for low-latency capture and speaker reference for echo cancellation.

Usage

This package is a pipeline node for @acpfx/cli. See the CLI package for installation and usage.

The postinstall script downloads a prebuilt binary for your platform. Supported: macOS (Apple Silicon), Linux (x86_64).

Manifest

• Consumes: audio.chunk (speaker reference for AEC), control.interrupt
• Emits: audio.chunk, audio.level, lifecycle.ready, lifecycle.done, control.error

Settings

| Name | Type | Default | Description | |------|------|---------|-------------| | sampleRate | number | 16000 | Target sample rate in Hz | | chunkMs | number | 100 | Chunk duration in milliseconds | | speaker | string | player | Node name whose audio to use as speaker reference for AEC | | debugDir | string | | Directory to write debug WAV recordings |

Pipeline Example

nodes:
  mic:
    use: "@acpfx/mic-speaker"
    settings: { sampleRate: 16000, speaker: player }
    outputs: [stt]
  player:
    use: "@acpfx/audio-player"
    settings: { speechSource: tts }
    outputs: [mic]  # cycle: reference audio for AEC

Building from Source

cargo build --release -p mic-speaker

Credits

Built on sys-voice, a Rust crate for native audio I/O with acoustic echo cancellation.

License

ISC