speech_recognition/FunASR
1
0
Fork 0
mirror of https://github.com/modelscope/FunASR synced 2025-09-15 14:48:36 +08:00
Code Issues Actions Packages Projects Releases Wiki Activity

Merge pull request #531 from alibaba-damo-academy/dev_new

Browse Source 
websocket online 2pass bugfix
This commit is contained in:
zhifu gao 2023-05-19 11:31:33 +08:00 committed by GitHub
commit 7eaf608c2d
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
4 changed files with 16 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
egs_modelscope/punctuation/punc_ct-transformer_zh-cn-common-vadrealtime-vocab272727/demo.py
View File

@ -9,7 +9,7 @@ logger.setLevel(logging.CRITICAL)
inference_pipeline = pipeline(
task=Tasks.punctuation,
model='damo/punc_ct-transformer_zh-cn-common-vad_realtime-vocab272727',
model_revision = 'v1.0.2'
model_revision='v1.0.2'
)
##################text二进制数据#####################

2
funasr/runtime/html5/readme_cn.md
View File

@ -28,7 +28,7 @@ Recorder
```shell
usage: h5Server.py [-h] [--host HOST] [--port PORT] [--certfile CERTFILE]
[--keyfile KEYFILE]
python h5Server.py --port 1337
python h5Server.py --port 1337 --keyfile server.key
```
## 2.启动ws or wss asr online srv
[具体请看online asr](https://github.com/alibaba-damo-academy/FunASR/tree/main/funasr/runtime/python/websocket)

17
funasr/runtime/python/websocket/ws_server_2pass.py
View File

@ -46,7 +46,7 @@ if args.punc_model != "":
inference_pipeline_punc = pipeline(
task=Tasks.punctuation,
model=args.punc_model,
model_revision=None,
model_revision="v1.0.2",
ngpu=args.ngpu,
ncpu=args.ncpu,
)
@ -74,6 +74,7 @@ async def ws_serve(websocket, path):
websocket.param_dict_punc = {'cache': list()}
websocket.vad_pre_idx = 0
speech_start = False
speech_end_i = False
websocket.wav_name = "microphone"
print("new user connected", flush=True)
@ -99,7 +100,9 @@ async def ws_serve(websocket, path):
# asr online
frames_asr_online.append(message)
if len(frames_asr_online) % websocket.chunk_interval == 0:
websocket.param_dict_asr_online["is_final"] = speech_end_i
if len(frames_asr_online) % websocket.chunk_interval == 0 or websocket.param_dict_asr_online["is_final"]:
audio_in = b"".join(frames_asr_online)
await async_asr_online(websocket, audio_in)
frames_asr_online = []
@ -115,12 +118,13 @@ async def ws_serve(websocket, path):
frames_asr.extend(frames_pre)
# asr punc offline
if speech_end_i or not websocket.is_speaking:
# print("vad end point")
audio_in = b"".join(frames_asr)
await async_asr(websocket, audio_in)
frames_asr = []
speech_start = False
frames_asr_online = []
websocket.param_dict_asr_online = {"cache": dict()}
# frames_asr_online = []
# websocket.param_dict_asr_online = {"cache": dict()}
if not websocket.is_speaking:
websocket.vad_pre_idx = 0
frames = []
@ -173,10 +177,13 @@ async def async_asr(websocket, audio_in):
async def async_asr_online(websocket, audio_in):
if len(audio_in) > 0:
audio_in = load_bytes(audio_in)
# print(websocket.param_dict_asr_online.get("is_final", False))
rec_result = inference_pipeline_asr_online(audio_in=audio_in,
param_dict=websocket.param_dict_asr_online)
# print(rec_result)
if websocket.param_dict_asr_online.get("is_final", False):
websocket.param_dict_asr_online["cache"] = dict()
return
# websocket.param_dict_asr_online["cache"] = dict()
if "text" in rec_result:
if rec_result["text"] != "sil" and rec_result["text"] != "waiting_for_more_voice":
# print("online", rec_result)

2
funasr/runtime/python/websocket/ws_server_online.py
View File

@ -106,8 +106,10 @@ async def ws_serve(websocket, path):
async def async_asr_online(websocket,audio_in):
if len(audio_in) >= 0:
audio_in = load_bytes(audio_in)
# print(websocket.param_dict_asr_online.get("is_final", False))
rec_result = inference_pipeline_asr_online(audio_in=audio_in,
param_dict=websocket.param_dict_asr_online)
# print(rec_result)
if websocket.param_dict_asr_online.get("is_final", False):
websocket.param_dict_asr_online["cache"] = dict()
if "text" in rec_result: