Speech recognition: transcribe the microphone or an audio file.
Synopsis
hear [-i file] [-l lang] [-T seconds] [-d deviceId] [--engine name]
Description
Listens on the microphone until you pause, then prints the transcript, modeled on the macOS hear CLI. Can also transcribe an existing audio file instead of listening live.
Options
-i file Transcribe an audio file (wav/mp3/ogg/webm) instead of listening; uses the enhanced on-device model (downloads it on first use).
-l lang BCP-47 language tag (default: auto-detect).
-T seconds Max listening time (default 30).
-d deviceId Microphone device id (see --devices); applies to the enhanced capture path only — the built-in recognizer always uses the system default input.
--engine name builtin | enhanced (default: enhanced when ready).
--devices List audio input devices.
--status Show the enhanced model state (downloading/ready + ETA).
--warmup Start the enhanced model download in the background.
Examples
$ hear
Listen on the microphone and print the transcript.
$ hear -i memo.wav
Transcribe an existing audio file.
$ hear --devices
List available microphones.
Notes
Requires microphone permission in the browser. The enhanced on-device model is downloaded on first use; hear-11 (ElevenLabs skill) is a separate, unrelated command despite the similar name.
See Also
say