diff --git a/backtalk/config.py b/backtalk/config.py
index d59f7f2..6b6571a 100644
--- a/backtalk/config.py
+++ b/backtalk/config.py
@@ -131,6 +131,8 @@
     # language pipeline (a=American, b=British, e/f/h/i/j/p/z = other
     # languages), so keep voice and accent matched.
     "voice": "bm_lewis",
+    # Device for local Kokoro TTS pipeline: "cpu" (default) or "cuda"
+    "tts_device": "cpu",
     # Speech recognition (faster-whisper, local, free).
     # Models: tiny.en / base.en / small.en / medium.en — small.en is the
     # accuracy/speed sweet spot on a normal machine.
diff --git a/backtalk/ears.py b/backtalk/ears.py
index ad52343..6cb9b66 100644
--- a/backtalk/ears.py
+++ b/backtalk/ears.py
@@ -289,12 +289,13 @@ def warm():
                                        verbose=None)
                 _model, _backend = repo, "mlx"
             else:
-                from faster_whisper import WhisperModel
                 want = CFG["stt_device"]
+                compute = CFG["stt_compute"]
                 log(f"[ears] loading {CFG['stt_model']} "
-                    f"({want}/{CFG['stt_compute']})...")
+                    f"({want}/{compute})...")
+                from faster_whisper import WhisperModel
                 _model = WhisperModel(CFG["stt_model"], device=want,
-                                      compute_type=CFG["stt_compute"])
+                                      compute_type=compute)
                 # PROVE the device before the greeting, not at the first
                 # spoken sentence. WhisperModel CONSTRUCTS perfectly well
                 # against a GPU it cannot actually use: "auto" picks CUDA
@@ -418,6 +419,12 @@ def record_held(is_held, max_s: float = 60.0, min_s: float = 0.25) -> str | None
             frames.append(block[:, 0].copy())
     if len(frames) * FRAME_MS / 1000 < min_s:
         return None
+    try:
+        from backtalk.signals import set_state
+        set_state("thinking")
+    except Exception:
+        pass
+    print("[ptt] transcribing...", flush=True)
     return transcribe(np.concatenate(frames))
 
 
diff --git a/backtalk/main.py b/backtalk/main.py
index a7b85d1..6403a81 100644
--- a/backtalk/main.py
+++ b/backtalk/main.py
@@ -679,7 +679,7 @@ async def amain():
     loop = asyncio.get_event_loop()
     # Warm the engines while the greeting plays: the STT model load and
     # the brain's prompt-cache toll both hide behind the spoken line.
-    loop.run_in_executor(None, warm_ears)
+    warm_ears_fut = loop.run_in_executor(None, warm_ears)
     # THE BRAIN CONNECT, guarded. This is the one startup step that
     # needs a signed-in Claude Code, internet, and available usage.
     # When it fails or hangs, the mouth still works, so SAY SO instead
@@ -707,6 +707,11 @@ async def _warmup():
         mouth.wait_done(timeout=30)
         raise SystemExit(1)
     log("[backtalk] brain warm")
+    try:
+        await asyncio.wait_for(asyncio.shield(warm_ears_fut), 60)
+        log("[backtalk] ears warm and ready")
+    except Exception as e:
+        log(f"[ears] warmup error: {e!r}")
     # the hidden warmup ping is plumbing, not conversation
     brain.session.update(turns=0, out_tokens=0, in_tokens=0, cost=0.0)
     # a configured effort level applies at launch (saved by the spoken
diff --git a/backtalk/mouth.py b/backtalk/mouth.py
index c3bc4f5..b7428f7 100644
--- a/backtalk/mouth.py
+++ b/backtalk/mouth.py
@@ -171,8 +171,9 @@ def warm():
             lang = (CFG["voice"] or "bm_lewis")[0]
             log(f"[mouth] loading kokoro (lang '{lang}', "
                 f"voice {CFG['voice']})...")
-            _pipe = KPipeline(lang_code=lang)
-            log("[mouth] voice ready")
+            device = CFG.get("tts_device", "cpu")
+            _pipe = KPipeline(lang_code=lang, device=device)
+            log(f"[mouth] voice ready ({device})")
     return _pipe