m3tam3re/nixos-config
1
0
Fork 0
Code Issues Pull Requests 1 Actions Packages Projects Releases Wiki Activity

fix: enable Hermes voice STT dependencies #21

Merged
m3tam3re merged 1 commits from fix/hermes-voice-stt-deps into master 2026-05-23 11:30:18 +02:00
Conversation 0 Commits 1 Files Changed 1
+2 -1
m3ta-chiron commented 2026-05-23 10:32:33 +02:00
Collaborator
Copy Link
Copy Source

Summary

  • Add the Hermes voice optional dependency group to m3-hermes
  • Keep the existing Matrix/Honcho/Exa/Edge TTS groups and update the comment to mention local STT
  • This bakes faster-whisper into the sealed Nix Hermes environment so Matrix voice messages can be transcribed without runtime pip/uv mutation

Validation

  • nix shell nixpkgs#alejandra -c alejandra hosts/m3-hermes/services/hermes-agent.nix
  • nix build .#nixosConfigurations.m3-hermes.config.system.build.toplevel --no-link --show-trace
  • Verified built Hermes Python reports feature_missing("stt.faster_whisper") == () and imports faster_whisper/numpy
  • Verified local STT transcribes the test OGG via tools.transcription_tools.transcribe_audio(..., model="tiny")

Note

  • sounddevice still cannot import standalone on NixOS because PortAudio is not discoverable via ctypes.util.find_library; this does not block Matrix voice-message transcription, which uses faster-whisper on cached audio files rather than microphone capture.
## Summary - Add the Hermes `voice` optional dependency group to `m3-hermes` - Keep the existing Matrix/Honcho/Exa/Edge TTS groups and update the comment to mention local STT - This bakes `faster-whisper` into the sealed Nix Hermes environment so Matrix voice messages can be transcribed without runtime pip/uv mutation ## Validation - `nix shell nixpkgs#alejandra -c alejandra hosts/m3-hermes/services/hermes-agent.nix` - `nix build .#nixosConfigurations.m3-hermes.config.system.build.toplevel --no-link --show-trace` - Verified built Hermes Python reports `feature_missing("stt.faster_whisper") == ()` and imports `faster_whisper`/`numpy` - Verified local STT transcribes the test OGG via `tools.transcription_tools.transcribe_audio(..., model="tiny")` ## Note - `sounddevice` still cannot import standalone on NixOS because PortAudio is not discoverable via `ctypes.util.find_library`; this does not block Matrix voice-message transcription, which uses `faster-whisper` on cached audio files rather than microphone capture.
m3ta-chiron added 1 commit 2026-05-23 10:32:34 +02:00
fix: enable Hermes voice STT dependencies 95aaddd8c0
m3tam3re merged commit 9f1e7fd568 into master 2026-05-23 11:30:18 +02:00
m3tam3re deleted branch fix/hermes-voice-stt-deps 2026-05-23 11:30:18 +02:00
Sign in to join this conversation.
Reviewers
No Reviewers
Labels
No items
No labels
Milestone
No items
No Milestone
Projects
Clear projects
No projects
Assignees
Clear assignees
m3ta-chiron m3tam3re
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: m3tam3re/nixos-config#21