fix: auto-submit always fires, plain Enter, TTS volume by hakanensari · Pull Request #4 · PerIPan/OpenWhisperer

hakanensari · 2026-03-10T13:03:28Z

Summary

Auto-submit always fires when enabled — no need to say "submit" or "send" at the end
Plain Enter instead of Cmd+Enter — fixes Ghostty maximizing the window. Explicitly clears modifier flags on the CGEvent so held Control key doesn't bleed into the Enter
TTS volume boost to 4x — Kokoro output is naturally quiet
mktemp guard — prevents poison temp files when mktemp fails
Removed outdated hint "say submit or send at end of phrase" from menubar UI

Test plan

Enable auto-submit, record a phrase, verify it auto-submits without a trigger word
Verify Ghostty does not maximize on auto-submit
Verify TTS playback is audible at comfortable volume

When setup runs a second time (e.g. app relaunch or resetAndRerun), `uv venv` fails because the venv directory already exists. This prevents .setup-complete from being written, so the Python server never starts. The --clear flag tells uv to replace an existing venv instead of erroring out.

Add design document for three interaction modes: press-to-talk (existing), hold-to-talk, and hands-free with silence detection and barge-in support.

Four-phase plan covering hold-to-talk, TTS latency optimization, hands-free mode with silence detection and barge-in, and polish. Phases 1 and 2 can run in parallel.

- Increase afplay volume to 4x (Kokoro output is quiet) - Guard mktemp failure to prevent poison temp files - Auto-submit fires on every transcription when enabled (no trigger word needed) - Send plain Enter instead of Cmd+Enter (Cmd+Enter maximizes Ghostty)

- Auto-submit fires on every transcription when enabled (no trigger word needed) - Send plain Enter with explicit zero modifier flags (fixes Ghostty maximizing on Ctrl+Enter) - Increase delay to 1s to ensure Control key is released before Enter - Boost afplay volume to 4x (Kokoro output is naturally quiet) - Guard mktemp failure to prevent poison temp files - Remove outdated "say submit or send" hint from menubar

Copilot

Pull request overview

This PR updates Claude Whisperer’s automation and speech pipeline to make auto-submit unconditional when enabled, adjust the synthetic keypress to plain Enter (avoiding modifier bleed), and improve TTS playback loudness and robustness.

Changes:

Auto-submit now always triggers when the auto-submit flag is enabled (no trigger phrase required).
Synthetic submit keypress changed from Cmd+Enter to plain Enter via CGEvent, explicitly clearing modifier flags and increasing the delay before pressing Enter.
TTS hook adds a mktemp failure guard and attempts a volume boost during afplay playback; menubar hint text is removed; docs for hands-free mode planning/design are added.

Reviewed changes

Copilot reviewed 6 out of 6 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
`servers/unified_server.py`	Switches to plain Enter key event + modifier clearing; makes auto-submit unconditional when enabled.
`hooks/tts-hook.sh`	Adds `mktemp` failure guard and changes playback invocation to include a volume flag.
`app/Sources/ClaudeWhisperer/SetupManager.swift`	Adjusts `uv venv` invocation to include `--clear`.
`app/Sources/ClaudeWhisperer/MenuBarView.swift`	Removes outdated “say submit/send” UI hint.
`docs/plans/2026-03-10-hands-free-mode-implementation.md`	Adds detailed hands-free/hold-to-talk implementation plan.
`docs/plans/2026-03-10-hands-free-mode-design.md`	Adds hands-free/hold-to-talk design doc.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

You can also share your feedback on Copilot code review. Take the survey.

Copilot · 2026-03-10T13:07:54Z

servers/unified_server.py

@@ -266,7 +267,8 @@ async def transcribe(

        should_submit = False
        if os.path.exists(AUTO_SUBMIT_FLAG):


With auto-submit enabled, should_submit is set to True unconditionally, even when the transcription result is empty/whitespace. This will still schedule a synthetic Enter keypress and can end up submitting an empty message. Consider gating auto-submit on text.strip() (or similar) so Enter is only sent when there’s something to submit.

Suggested change

if os.path.exists(AUTO_SUBMIT_FLAG):

# Only auto-submit when there is non-whitespace transcription text.

if os.path.exists(AUTO_SUBMIT_FLAG) and text.strip():

Copilot · 2026-03-10T13:07:54Z

servers/unified_server.py

        if os.path.exists(AUTO_SUBMIT_FLAG):
-            text, should_submit = check_submit_trigger(text)
+            text, _ = check_submit_trigger(text)
+            should_submit = True


Now that auto-submit always fires, calling check_submit_trigger() solely to strip trailing trigger phrases will remove legitimate dictated content whenever it ends with any entry in SUBMIT_TRIGGERS (e.g., "enter", "go ahead", "send"). If the goal is just to drop the legacy "submit/send" habit, consider narrowing the strip list (or making stripping optional) so normal language endings aren’t silently deleted.

Copilot · 2026-03-10T13:07:55Z

app/Sources/ClaudeWhisperer/SetupManager.swift

            guard runCommand(
                Paths.uvBinary.path,
-                args: ["venv", Paths.venv.path, "--python", "3.13"],
+                args: ["venv", Paths.venv.path, "--python", "3.13", "--clear"],
                step: "Creating Python environment..."


uv venv is invoked with --clear. The repo’s setup.sh uses uv venv ... --python 3.13 without this flag, and not all uv versions support --clear. If the bundled uv doesn’t recognize it, first-launch setup will fail. Consider explicitly deleting Paths.venv when re-creating the environment (or probing supported flags / uv version) instead of relying on --clear.

Copilot · 2026-03-10T13:07:55Z

hooks/tts-hook.sh


  if [ -s "$TMPFILE" ]; then
-    afplay "$TMPFILE" 2>/dev/null
+    afplay -v 4 "$TMPFILE" 2>/dev/null


afplay -v expects a normalized volume scalar (typically 0.0–1.0). Passing -v 4 may be clamped or cause playback to fail, which would undermine the intended volume boost. Consider using a supported range for -v and, if you need a true 4× gain, amplify/normalize the WAV data before playback or make the gain configurable with a safe ceiling.

Suggested change

afplay -v 4 "$TMPFILE" 2>/dev/null

# Use configurable volume, clamped to afplay's expected 0.0–1.0 range

RAW_VOLUME="${TTS_VOLUME:-1.0}"

VOLUME=$(awk -v v="$RAW_VOLUME" 'BEGIN { if (v == "" || v < 0) print 1.0; else if (v > 1.0) print 1.0; else print v }')

afplay -v "$VOLUME" "$TMPFILE" 2>/dev/null

Rename all refs, bundle ID, App Support path, source dir. Apply PR #1 (uv venv --clear), PR #4 (plain Enter, TTS vol boost, auto-submit always fires), PR #5 (mktemp hardening). Fix .gitignore to track source files.

hakanensari added 6 commits March 10, 2026 11:22

design: hands-free voice mode

bf3ae44

Add design document for three interaction modes: press-to-talk (existing), hold-to-talk, and hands-free with silence detection and barge-in support.

plan: hands-free mode implementation steps

79bd47b

Four-phase plan covering hold-to-talk, TTS latency optimization, hands-free mode with silence detection and barge-in, and polish. Phases 1 and 2 can run in parallel.

Merge branch 'design/hands-free-mode'

c54534b

Copilot AI review requested due to automatic review settings March 10, 2026 13:03

Copilot started reviewing on behalf of hakanensari March 10, 2026 13:03 View session

Copilot AI reviewed Mar 10, 2026

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

fix: auto-submit always fires, plain Enter, TTS volume#4

fix: auto-submit always fires, plain Enter, TTS volume#4
hakanensari wants to merge 6 commits intoPerIPan:mainfrom
hakanensari:fix/auto-submit-improvements

hakanensari commented Mar 10, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Mar 10, 2026

Uh oh!

Copilot AI Mar 10, 2026

Uh oh!

Copilot AI Mar 10, 2026

Uh oh!

Copilot AI Mar 10, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

		@@ -266,7 +267,8 @@ async def transcribe(

		should_submit = False
		if os.path.exists(AUTO_SUBMIT_FLAG):

	if os.path.exists(AUTO_SUBMIT_FLAG):
	# Only auto-submit when there is non-whitespace transcription text.
	if os.path.exists(AUTO_SUBMIT_FLAG) and text.strip():

-    afplay -v 4 "$TMPFILE" 2>/dev/null
+    # Use configurable volume, clamped to afplay's expected 0.0–1.0 range
+    RAW_VOLUME="${TTS_VOLUME:-1.0}"
+    VOLUME=$(awk -v v="$RAW_VOLUME" 'BEGIN { if (v == "" || v < 0) print 1.0; else if (v > 1.0) print 1.0; else print v }')
+    afplay -v "$VOLUME" "$TMPFILE" 2>/dev/null

Conversation

hakanensari commented Mar 10, 2026

Summary

Test plan

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Mar 10, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 10, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 10, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 10, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants