 wenhongquan
|
baa00cd81f
fix: change dispatcher port to 9001 (9000 occupied)
|
2 月之前 |
 wenhongquan
|
d1afb91175
build(android): 配置安卓签名和混淆规则，更新默认连接地址
|
2 月之前 |
|
wenhongquan
|
29e3519f87
feat(dispatcher): multi-user room management service
|
2 月之前 |
|
wenhongquan
|
0b3134495e
fix(tts+asr): parallel TTS with ordered enqueue + filter single-char ASR filler
|
2 月之前 |
|
wenhongquan
|
34cb936a5e
fix(tts): ordered TTS generation - collect segments first, generate sequentially
|
2 月之前 |
|
wenhongquan
|
c80cfcdb22
fix(tts): keep audio int16, fix indentation
|
2 月之前 |
|
wenhongquan
|
cf28ff0576
fix(vad): model path aligned with Dockerfile (whisper_asr/silero_vad.onnx)
|
2 月之前 |
|
wenhongquan
|
4b1a2ae8db
fix: restore _send method, move _drain_queue before __main__
|
2 月之前 |
|
wenhongquan
|
beea2f5be1
fix: move _drain_queue to module level after Worker class
|
2 月之前 |
|
wenhongquan
|
f648952a0a
fix: indent _drain_queue and _send inside Worker class
|
2 月之前 |
|
wenhongquan
|
7a5790c994
refactor(tts): single persistent track + queue playback + gen-based interrupt
|
2 月之前 |
|
wenhongquan
|
ca38fff29c
fix(asr): concatenate streaming ASR chunks instead of keeping only last
|
2 月之前 |
|
wenhongquan
|
18e3e4d341
fix(flutter): display streaming AI reply (liveReply) in conversation list
|
2 月之前 |
|
wenhongquan
|
1ba163922d
feat(tts-asr): streaming Mimo TTS (voiceclone) + streaming ASR
|
2 月之前 |
|
wenhongquan
|
a381dc1067
feat(provider): add LLM/ASR provider switch (vllm/mimo, qwen/mimo)
|
2 月之前 |
|
wenhongquan
|
fbe9284c3e
revert: remove partial ASR (CPU thrashing hurt latency)
|
2 月之前 |
|
wenhongquan
|
3d398e2baf
build(asr_agent): add onnxruntime dependency
|
2 月之前 |
|
wenhongquan
|
b3de0309a2
perf(asr): streaming partial ASR during speech + reduce VAD silence to 0.4s
|
2 月之前 |
|
wenhongquan
|
6afa008037
feat(tts): concurrent TTS playback + user interruption (xiaozhi-style abort)
|
2 月之前 |
|
wenhongquan
|
57e325ab76
feat(vad): bundle Silero VAD ONNX model (2.3MB) in asr_agent/
|
2 月之前 |
|
wenhongquan
|
5cec1bb5ae
chore: add .gitignore
|
2 月之前 |
|
wenhongquan
|
4ffc9e9139
feat(flutter): add ReplyPartialMessage to handle streaming LLM replies
|
2 月之前 |
|
wenhongquan
|
7a424e632b
feat(vad): replace energy-based VAD with ONNX Silero VAD (xiaozhi-style)
|
2 月之前 |
|
wenhongquan
|
9788bcb353
feat(tts): xiaozhi-style two-tier sentence segmentation + markdown cleaner + think tag filter
|
2 月之前 |
|
wenhongquan
|
ea42df876d
feat(llm-tts): streaming LLM + incremental TTS segmentation (xiaozhi pattern)
|
2 月之前 |
|
wenhongquan
|
2478295a00
fix(llm): use LLM_MODEL env var instead of hardcoded 'default'
|
2 月之前 |
|
wenhongquan
|
1e97164777
rewrite(conversation-worker): full pipeline rewrite based on xiaozhi-esp32-server patterns
|
2 月之前 |
|
wenhongquan
|
1059df76bc
fix(worker): use non-overlapping transcription window to eliminate duplicate outputs
|
2 月之前 |
|
wenhongquan
|
a221c1ae02
fix(livekit): add participant_connected and track_published handlers to fix track_subscribed not firing
|
2 月之前 |
|
wenhongquan
|
cee917bf35
refactor(transcript-proc): 重构转录处理器，支持批量返回处理结果
|
2 月之前 |
