feat: ack tone, duck-before-TTS, instant ducking on voice/unmute

- Add ascending two-tone chime (880Hz/1320Hz) before TTS playback
  as audible acknowledgment that the voice trigger was recognized
- Signal music ducking 1.5s before TTS starts so music is already
  lowered when audio begins playing
- Snap duck volume to floor instantly on voice packet or user unmute
  via pymumble callback, eliminating the 1s poll delay
- Register USERUPDATED callback to preemptively duck when a user
  unmutes (they're about to speak)
- Strip leading punctuation from trigger remainder (Whisper artifacts)

plugins/voice.py

@@ -11,6 +11,8 @@ import asyncio
 import io
 import json
 import logging
+import math
+import struct
 import threading
 import time
 import urllib.request
@@ -83,6 +85,50 @@ def _pcm_to_wav(pcm: bytes) -> bytes:
     return buf.getvalue()
 
 
+# -- Acknowledge tone --------------------------------------------------------
+
+_ACK_FREQ = (880, 1320)    # A5 -> E6 ascending
+_ACK_NOTE_DUR = 0.15        # seconds per note
+_ACK_AMP = 12000            # gentle amplitude
+_ACK_FRAME = 960            # 20ms at 48kHz, matches Mumble native
+
+
+async def _ack_tone(bot) -> None:
+    """Play a short two-tone ascending chime via pymumble sound_output."""
+    mu = getattr(bot, "_mumble", None)
+    if mu is None:
+        return
+    so = mu.sound_output
+    if so is None:
+        return
+
+    # Unmute if self-muted (stream_audio handles re-mute later)
+    if getattr(bot, "_self_mute", False):
+        if bot._mute_task and not bot._mute_task.done():
+            bot._mute_task.cancel()
+            bot._mute_task = None
+        try:
+            mu.users.myself.unmute()
+        except Exception:
+            pass
+
+    frames_per_note = int(_ACK_NOTE_DUR / 0.02)  # 0.02s per frame
+    for freq in _ACK_FREQ:
+        for i in range(frames_per_note):
+            samples = []
+            for j in range(_ACK_FRAME):
+                t = (i * _ACK_FRAME + j) / _SAMPLE_RATE
+                samples.append(int(_ACK_AMP * math.sin(2 * math.pi * freq * t)))
+            pcm = struct.pack(f"<{_ACK_FRAME}h", *samples)
+            so.add_sound(pcm)
+            while so.get_buffer_size() > 0.5:
+                await asyncio.sleep(0.02)
+
+    # Wait for tone to finish
+    while so.get_buffer_size() > 0:
+        await asyncio.sleep(0.05)
+
+
 # -- STT: Sound listener (pymumble thread) ----------------------------------
 
 
@@ -170,7 +216,7 @@ async def _flush_monitor(bot):
 
                 trigger = ps["trigger"]
                 if trigger and text.lower().startswith(trigger.lower()):
-                    remainder = text[len(trigger):].strip()
+                    remainder = text[len(trigger):].strip().lstrip(",.;:!?")
                     if remainder:
                         log.info("voice: trigger from %s: %s", name, remainder)
                         bot._spawn(
@@ -243,8 +289,10 @@ async def _tts_play(bot, text: str):
     if wav_path is None:
         return
     try:
-        # Signal music plugin to duck while TTS is playing
+        # Signal music plugin to duck, wait for it to take effect
         bot.registry._tts_active = True
+        await asyncio.sleep(1.5)
+        await _ack_tone(bot)
         done = asyncio.Event()
         await bot.stream_audio(str(wav_path), volume=1.0, on_done=done)
         await done.wait()

src/derp/mumble.py

@@ -20,6 +20,7 @@ from pymumble_py3.constants import (
     PYMUMBLE_CLBK_DISCONNECTED,
     PYMUMBLE_CLBK_SOUNDRECEIVED,
     PYMUMBLE_CLBK_TEXTMESSAGERECEIVED,
+    PYMUMBLE_CLBK_USERUPDATED,
 )
 
 from derp.bot import _TokenBucket
@@ -217,6 +218,10 @@ class MumbleBot:
             PYMUMBLE_CLBK_SOUNDRECEIVED,
             self._on_sound_received,
         )
+        self._mumble.callbacks.set_callback(
+            PYMUMBLE_CLBK_USERUPDATED,
+            self._on_user_updated,
+        )
         self._mumble.set_receive_sound(self._receive_sound)
         self._mumble.start()
         self._mumble.is_ready()
@@ -287,6 +292,18 @@ class MumbleBot:
         log.warning("mumble: disconnected")
         self._last_voice_ts = 0.0
 
+    def _instant_duck(self) -> None:
+        """Snap music volume to duck floor immediately.
+
+        Called from pymumble thread on voice/unmute events so ducking
+        takes effect on the next audio frame (~20ms) instead of waiting
+        for the 1s duck monitor poll.
+        """
+        for peer in getattr(self.registry, "_bots", {}).values():
+            ps = getattr(peer, "_pstate", {}).get("music")
+            if ps and ps.get("duck_enabled") and ps.get("task"):
+                ps["duck_vol"] = float(ps["duck_floor"])
+
     def _on_sound_received(self, user, sound_chunk) -> None:
         """Callback from pymumble thread: voice audio received.
 
@@ -302,12 +319,33 @@ class MumbleBot:
         self.registry._voice_ts = self._last_voice_ts
         if prev == 0.0:
             log.info("mumble: first voice packet from %s", name or "?")
+        self._instant_duck()
         for fn in self._sound_listeners:
             try:
                 fn(user, sound_chunk)
             except Exception:
                 log.exception("mumble: sound listener error")
 
+    def _on_user_updated(self, user, actions) -> None:
+        """Callback from pymumble thread: user state changed.
+
+        When a non-bot user unmutes, update ``_voice_ts`` and snap duck
+        volume to floor immediately.
+        """
+        if "self_mute" not in actions:
+            return
+        name = user["name"] if isinstance(user, dict) else None
+        bots = getattr(self.registry, "_bots", {})
+        if name and name in bots:
+            return
+        # Only care about unmute (self_mute going False)
+        if user.get("self_mute", True):
+            return
+        log.info("mumble: %s unmuted, preemptive duck", name or "?")
+        self._last_voice_ts = time.monotonic()
+        self.registry._voice_ts = self._last_voice_ts
+        self._instant_duck()
+
     def _on_text_message(self, message) -> None:
         """Callback from pymumble thread: text message received.