Skip to content

Hands-free voice mode (VAD) + install-on-demand for voice dependencies (PR4 of 6) - #797

Merged
will-lamerton merged 12 commits into
Nano-Collective:feature/voice-modefrom
RONAK-AI647:RONAK-AI647/PR4-Hands-Free-VAD
Aug 9, 2026
Merged

Hands-free voice mode (VAD) + install-on-demand for voice dependencies (PR4 of 6)#797
will-lamerton merged 12 commits into
Nano-Collective:feature/voice-modefrom
RONAK-AI647:RONAK-AI647/PR4-Hands-Free-VAD

Conversation

@RONAK-AI647

Copy link
Copy Markdown
Contributor

What this does

Two things in this PR:

  1. No more manual terminal installs. The first time you try voice mode and
    the required tools (sox, whisper, piper) aren't installed, Nanocoder now asks
    you right there in the app - "Install now? Y/N" - and installs everything
    automatically if you say yes. No separate terminal, no manual steps.

  2. Real hands-free voice. You can now switch to /voice hands-free and just
    talk - no need to hold a button. Nanocoder listens continuously and
    automatically detects when you start and stop speaking.

Known limitation

The hands-free listening uses a simple volume-based detection method rather than
a full AI speech detector. This keeps it lightweight and working across every
platform, but it means loud background noise could occasionally be picked up as
speech. This is a reasonable first version - a smarter version can come later if
needed.

Test 1: Install prompt

  1. Make sure sox/whisper/piper are NOT installed.
  2. Run /voice or Ctrl+T.
  3. Confirm you see: "Voice mode requires sox, whisper, piper (~150 MB). Install now?"
  4. Press Esc/No - confirm it cancels cleanly and doesn't ask again this session.
  5. Restart the app, trigger voice again - confirm it DOES ask again (session-only memory).

Test 2: Real install

  1. Trigger the prompt again, select Yes.
  2. Confirm you see actual progress updates, not just a frozen "Installing..." screen.
  3. Once done, run the same command-line check we used before
    (e.g. which sox, which whisper-cli, which piper) and confirm all three
    are genuinely present - not just sox.

Test 3: Hands-free mode

  1. Run /voice hands-free.
  2. Speak without pressing anything - confirm the status bar shows Listening
    automatically.
  3. Stop speaking - confirm it transitions to Processing, then Speaking (reads
    the reply), then back to Listening automatically for the next turn.

Test 4: Push-to-talk still works

  1. Run /voice push-to-talk.
  2. Confirm Ctrl+T still works exactly as it did in PR3.

@will-lamerton

Copy link
Copy Markdown
Member

Nice work @RONAK-AI647 :)

@will-lamerton
will-lamerton merged commit 48314bd into Nano-Collective:feature/voice-mode Aug 9, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants