Skip to content

Open-source speech recognition toolkit for training, inference, streaming ASR, VAD, punctuation, speaker diarization pipelines, and OpenAI-compatible/MCP serving.

refactor: move DynamicStreamingVAD to funasr/models/fsmn_vad_streaming/

xiaoyunchong.xyc7be505f00ed0
diff --git a/docs/vllm_guide.md b/docs/vllm_guide.md
index f50cc8908..62950051a 100644
--- a/docs/vllm_guide.md
+++ b/docs/vllm_guide.md
@@ -517,7 +517,7 @@ Chrome 要求 HTTPS 或 localhost。远程服务器用 SSH 端口转发:`ssh -
## 附录:DynamicStreamingVAD
-`funasr.utils.dynamic_vad.DynamicStreamingVAD` 是通用的动态阈值流式 VAD 封装,
+`funasr.models.fsmn_vad_streaming.dynamic_vad.DynamicStreamingVAD` 是通用的动态阈值流式 VAD 封装,
在 fsmn-vad 基础上根据当前语音段的累积时长动态调整静音切分阈值。
### 设计动机
@@ -530,7 +530,7 @@ fsmn-vad 默认使用固定静音阈值(800ms)。实际场景中:
```python
from funasr import AutoModel
-from funasr.utils.dynamic_vad import DynamicStreamingVAD
+from funasr.models.fsmn_vad_streaming.dynamic_vad import DynamicStreamingVAD
vad_model = AutoModel(model="fsmn-vad", device="cuda:0")
diff --git a/examples/industrial_data_pretraining/fun_asr_nano/serve_realtime_ws.py b/examples/industrial_data_pretraining/fun_asr_nano/serve_realtime_ws.py
index a54e047bb..5c58bc5ab 100644
--- a/examples/industrial_data_pretraining/fun_asr_nano/serve_realtime_ws.py
+++ b/examples/industrial_data_pretraining/fun_asr_nano/serve_realtime_ws.py
@@ -71,7 +71,7 @@ def _clean_asr_text(text):
return text.strip()
-from funasr.utils.dynamic_vad import DynamicStreamingVAD
+from funasr.models.fsmn_vad_streaming.dynamic_vad import DynamicStreamingVAD
class HybridSpeakerTracker:
diff --git a/funasr/utils/dynamic_vad.py b/funasr/models/fsmn_vad_streaming/dynamic_vad.py
similarity index 98%
rename from funasr/utils/dynamic_vad.py
rename to funasr/models/fsmn_vad_streaming/dynamic_vad.py
index 35df65ac4..bd14515c9 100644
--- a/funasr/utils/dynamic_vad.py
+++ b/funasr/models/fsmn_vad_streaming/dynamic_vad.py
@@ -8,7 +8,7 @@
Usage (流式):
from funasr import AutoModel
- from funasr.utils.dynamic_vad import DynamicStreamingVAD
+ from funasr.models.fsmn_vad_streaming.dynamic_vad import DynamicStreamingVAD
vad_model = AutoModel(model="fsmn-vad", device="cuda:0")
vad = DynamicStreamingVAD(vad_model)