omp packages

@luxass/pi-voice0.3.3

Record voice input with /voice, transcribe locally or through an API, and paste it into Pi's composer.

GitHub Actions Published 8 days ago MIT ~1.3K downloads/mo

Untested on omp

This is a pi package. No one has reported whether it works on omp yet. If you try it, add your result below.

Did it work for you?

Result

One report per person. A new report replaces your old one.

Reports 0

No reports yet.

Readme

@luxass/pi-voice

Record speech and paste the transcription into Pi's composer.

Install

pi install npm:@luxass/pi-voice

Requires Node.js 24+ and microphone access. Linux also needs libasound.so.2. Run /voice setup to choose a local or API transcription profile and an input device. Local transcription uses GGUF models; the model picker can download Whisper or Parakeet models.

Use

Press Ctrl+Shift+V or run /voice to start recording. Requirements are checked before the microphone starts. If something is missing, Voice explains it and offers to open setup. Completing setup does not start recording; press the shortcut or run /voice again when you are ready. Press it again to stop, transcribe, and paste the result into the composer. The extension does not send your message for you.

Command Purpose
/voice Start or stop recording and transcribe on stop.
/voice cancel Discard the current recording.
/voice setup Choose a profile and input, then check requirements.
/voice device Choose an input device or the system default.
/voice profile add Create a local or API transcription profile.
/voice profile Switch between saved profiles.
/voice profile model Change the active profile's model.
/voice status Show the current profile and input device.
/voice doctor Check recording and transcription setup.

A local profile without an explicit model uses the first GGUF model in ~/.cache/agent-voice. Set AGENT_VOICE_MODEL_DIR to keep models elsewhere. For API transcription, /voice profile add asks for the endpoint, model, request format, and name of an environment variable holding the API key. Keep the key in your environment, not in the settings file.

Settings are loaded once at boot from Pi's agent directory, using voice-settings.json. Set AGENT_VOICE_SETTINGS to use a different file. TypeBox validates known fields, while extra keys are accepted without warnings. If the file is missing, Voice shows a setup hint. If it cannot be loaded or validated, Voice shows the errors and keeps /voice setup available. Loading never rewrites the file. Setup saves a complete configuration and updates it in memory; manual file edits require /reload. A recording keeps the profile it started with.