Fix mic drop issues

<details>
<summary>Prompts Used</summary>
Investigate and fix the bug in https://github.com/jimrandomh/faceclaw/issues/18

> Fixed. Here's what was going on and what changed.
>
> ## Root cause
>
> The issue's diagnosis was right, and the underlying flaw is a bit broader than the holder set: **`captureHolders` is phone-side bookkeeping, but the mic enable lives in the glasses' EvenHub session.** That enable dies — silently, from the phone's point of view — on a transport drop, the charging case, an EvenHub screen-off suspend, or a refused enable (`startG2AudioCapture` returns false when the display path isn't warm, and the Java worker then exits while the JS side still believes it started).
>
> `acquireCapture()` answered "is the mic running?" from the holder set, so a single stranded holder (a Transcribe window that was open when the link dropped) made every later request — wakeword assistant included — return early onto a stream that no longer existed, for the life of the process.
>
> The reported repro hits it via `resumeEvenHubSession()`, which resets `audioCaptureActive` and replays the prelude without re-enabling the mic; nothing on either side cleaned up.
>
> ## The fix
>
> **Java** — expose the real state of the stream instead of inferring it:
> - `FaceclawBleCommunicator.isAudioCaptureActive()` (`App_Resources/.../FaceclawBleCommunicator.java:363`) — true only while the session is up, not shut down, and the enable has been acked. Every teardown path already clears that flag.
> - `FaceclawVoiceController.isCapturing()` — `started` plus the communicator's view, with a lock-guarded `audioStarted` so the mic-warmup window doesn't read as dead.
>
> **`app/native/voice-control.ts`** — the bridge now asks `micIsLive()` rather than trusting the set, and tears a dead capture down before starting a new one. Added `handleSessionEnded()` (parks live holders, drops the local state) and `resumeCapture()` (restarts them on the next session); `releaseCapture`/`stop` clear the parked set. The raw-PCM tap for EvenHub mic apps had the identical staleness bug (`if (this.rawActive) return true`) and is covered the same way.
>
> **`app/g2/dashboard-controller.ts`** — wired the two session boundaries: any phase leaving `connected` parks the capture, and the first composited frame of the next session resumes it (a rendered frame is exactly the signal that a mic enable will be accepted). The screen-off EvenHub suspend is now deferred while a capture is held, so the screen timeout no longer kills a live transcription; a capture that races the suspend is parked by the same path.
>
> So Transcribe recovers on its own after a reconnect instead of sitting on a dead "Listening…", and the wakeword assistant is no longer blocked by anything left behind.
>
> ## Verification
>
> `./build.sh` (Java + webpack) clean, `tsc --noEmit` clean, `npm test` 61/61. Not committed.
>
> The behavior is all glasses-session state, which the test harness (pure modules only) can't reach, so it needs hardware. I wrote `notes/voice-capture-session-resume.md` with the five checks that matter — most importantly: Transcribe open → walk out of range → return → "Hey Even" should now get audio, and Transcribe should resume itself.
</details>
This commit is contained in:
James Babcock committed 2026-08-29 10:16:52 -07:00
1 parent fe28a6de4b
commit 756ee4e63b
4 files changed
+188 -15

No files matched your search

@@ -360,6 +360,19 @@ public class FaceclawBleCommunicator implements FaceclawBleListener, Runnable {
}
}
/**
* Whether the glasses mic is enabled right now. The enable lives in the
* current EvenHub session, so it dies with a transport drop, the charging
* case, or a suspend — silently, from the phone's point of view. Callers
* that track a capture across those events must check this rather than
* assume their earlier enable still holds.
*/
public boolean isAudioCaptureActive() {
synchronized (lock) {
return running && sessionReady && !shutdownRequested && audioCaptureActive;
}
}
/**
* Acquire/renew or release CFW's volatile wake-takeover lease on both
* arms. Delivery (not a protocol ACK) is awaited so a caller can ensure
@@ -76,6 +76,9 @@ public class FaceclawVoiceController {
private volatile FaceclawBleCommunicator communicator;
private Thread workerThread;
private volatile boolean started;
// Set once the worker has the glasses mic enabled for this session.
// Read and written under `lock`, so it flips with `started` atomically.
private boolean audioStarted;
private VoiceInputMode mode = VoiceInputMode.CLOUD;
private OfflineRecognizer recognizer;
private FaceclawLc3Decoder lc3Decoder;
@@ -137,11 +140,36 @@ public class FaceclawVoiceController {
}
mode = parseMode(requestedMode);
started = true;
audioStarted = false;
workerThread = new Thread(this::runLoop, "FaceclawVoiceController");
workerThread.start();
}
}
/**
* Whether mic audio is actually flowing. {@link #start} only records
* intent: the enable lives in the glasses' EvenHub session, so a transport
* drop or a session suspend can leave this controller started with a
* worker that will never see another packet. Anything deciding whether to
* (re)start capture must ask this rather than assume its own bookkeeping.
*/
public boolean isCapturing() {
boolean audioUp;
synchronized (lock) {
if (!started) {
return false;
}
audioUp = audioStarted;
}
if (!audioUp) {
// The worker is still bringing the mic up; report it as running so
// a concurrent request shares it instead of restarting it.
return true;
}
FaceclawBleCommunicator currentCommunicator = communicator;
return currentCommunicator != null && currentCommunicator.isAudioCaptureActive();
}
public void stop() {
Thread threadToJoin;
synchronized (lock) {
@@ -200,6 +228,9 @@ public class FaceclawVoiceController {
emitStatus("Could not start G2 microphone input.");
return;
}
synchronized (lock) {
audioStarted = true;
}
emitStatus(currentMode == VoiceInputMode.CLOUD
? "Listening (cloud)..."
@@ -220,6 +251,7 @@ public class FaceclawVoiceController {
releaseLc3();
synchronized (lock) {
started = false;
audioStarted = false;
workerThread = null;
}
}