ESP32 ignores binary audio unless it receives tts start first.
Also skip silent frames to reduce bandwidth.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
ESP32 stops sending audio when it receives tts start because it
thinks the server is speaking. Just acknowledge detect silently.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Root cause: Frida crashed because open_voice triggered UI transition
while hook was active. Now: open voice chat first (one-shot script),
then attach bridge when libantaudio.so is already loaded.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Root cause: recv_loop started before open_voice completed, bridge
connection died during UI transition. Now setup completes first.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
hailin