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 #468 from alibaba-damo-academy/dev_websocket_offline

Browse Source 
Dev websocket offline
This commit is contained in:
zhifu gao 2023-05-08 00:13:59 +08:00 committed by GitHub
commit 7557fa87a8
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
5 changed files with 31 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
funasr/runtime/python/websocket/README.md
View File

@ -58,7 +58,7 @@ python ws_client.py --host "0.0.0.0" --port 10095 --chunk_interval 10 --words_ma
##### Loadding from wav.scp(kaldi style)
```shell
# --chunk_interval, "10": 600/10=60ms, "5"=600/5=120ms, "20": 600/12=30ms
python ws_client.py --host "0.0.0.0" --port 10095 --chunk_interval 10 --words_max_print 100 --audio_in "./data/wav.scp" --send_without_sleep
python ws_client.py --host "0.0.0.0" --port 10095 --chunk_interval 10 --words_max_print 100 --audio_in "./data/wav.scp" --send_without_sleep --output_dir "./results"
```
#### ASR streaming client
##### Recording from mircrophone
@ -69,7 +69,7 @@ python ws_client.py --host "0.0.0.0" --port 10095 --chunk_size "5,10,5" --words_
##### Loadding from wav.scp(kaldi style)
```shell
# --chunk_size, "5,10,5"=600ms, "8,8,4"=480ms
python ws_client.py --host "0.0.0.0" --port 10095 --chunk_size "5,10,5" --audio_in "./data/wav.scp" --words_max_print 100
python ws_client.py --host "0.0.0.0" --port 10095 --chunk_size "5,10,5" --audio_in "./data/wav.scp" --words_max_print 100 --output_dir "./results"
```
#### ASR offline/online 2pass client

23
funasr/runtime/python/websocket/ws_client.py
View File

@ -7,7 +7,9 @@ import asyncio
import argparse
import json
import traceback
from multiprocessing import Process
from multiprocessing import Process
from funasr.fileio.datadir_writer import DatadirWriter
parser = argparse.ArgumentParser()
parser.add_argument("--host",
type=str,
@ -43,6 +45,10 @@ parser.add_argument("--words_max_print",
type=int,
default=100,
help="chunk")
parser.add_argument("--output_dir",
type=str,
default=None,
help="output_dir")
args = parser.parse_args()
args.chunk_size = [int(x) for x in args.chunk_size.split(",")]
@ -51,6 +57,11 @@ print(args)
from queue import Queue
voices = Queue()
ibest_writer = None
if args.output_dir is not None:
writer = DatadirWriter(args.output_dir)
ibest_writer = writer[f"1best_recog"]
async def record_microphone():
is_finished = False
import pyaudio
@ -91,7 +102,9 @@ async def record_from_scp():
wavs = [args.audio_in]
for wav in wavs:
wav_splits = wav.strip().split()
wav_name = wav_splits[0] if len(wav_splits) > 1 else "demo"
wav_path = wav_splits[1] if len(wav_splits) > 1 else wav_splits[0]
# bytes_f = open(wav_path, "rb")
# bytes_data = bytes_f.read()
with wave.open(wav_path, "rb") as wav_file:
@ -112,7 +125,7 @@ async def record_from_scp():
beg = i*stride
data = audio_bytes[beg:beg+stride]
data = data.decode('ISO-8859-1')
message = json.dumps({"chunk_size": args.chunk_size, "chunk_interval": args.chunk_interval, "is_speaking": is_speaking, "audio": data, "is_finished": is_finished})
message = json.dumps({"chunk_size": args.chunk_size, "chunk_interval": args.chunk_interval, "is_speaking": is_speaking, "audio": data, "is_finished": is_finished, "wav_name": wav_name})
voices.put(message)
# print("data_chunk: ", len(data_chunk))
# print(voices.qsize())
@ -152,14 +165,18 @@ async def message(id):
# print(meg, end = '')
# print("\r")
# print(meg)
wav_name = meg.get("wav_name", "demo")
print(wav_name)
text = meg["text"][0]
if ibest_writer is not None:
ibest_writer["text"][wav_name] = text
if meg["mode"] == "online":
text_print += " {}".format(text)
else:
text_print += "{}".format(text)
text_print = text_print[-args.words_max_print:]
os.system('clear')
print("\r"+str(id)+":"+text_print)
print("\rpid"+str(id)+": "+text_print)
except Exception as e:
print("Exception:", e)
traceback.print_exc()

6
funasr/runtime/python/websocket/ws_server_offline.py
View File

@ -78,6 +78,7 @@ async def ws_serve(websocket, path):
is_speaking = message["is_speaking"]
websocket.param_dict_vad["is_final"] = not is_speaking
websocket.wav_name = message.get("wav_name", "demo")
if speech_start:
frames_asr.append(audio)
speech_start_i, speech_end_i = await async_vad(websocket, audio)
@ -136,8 +137,9 @@ async def async_asr(websocket, audio_in):
rec_result = inference_pipeline_punc(text_in=rec_result['text'],
param_dict=websocket.param_dict_punc)
# print(rec_result)
message = json.dumps({"mode": "offline", "text": [rec_result["text"]]})
await websocket.send(message)
message = json.dumps({"mode": "offline", "text": [rec_result["text"]], "wav_name": websocket.wav_name})
await websocket.send(message)

8
funasr/runtime/python/websocket/ws_server_online.py
View File

@ -53,7 +53,7 @@ async def ws_serve(websocket, path):
is_speaking = message["is_speaking"]
websocket.param_dict_asr_online["is_final"] = not is_speaking
websocket.wav_name = message.get("wav_name", "demo")
websocket.param_dict_asr_online["chunk_size"] = message["chunk_size"]
frames_online.append(audio)
@ -81,9 +81,9 @@ async def async_asr_online(websocket,audio_in):
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":
if len(rec_result["text"])>0:
rec_result["text"][0]=rec_result["text"][0] #.replace(" ","")
message = json.dumps({"mode": "online", "text": rec_result["text"]})
# if len(rec_result["text"])>0:
# rec_result["text"][0]=rec_result["text"][0] #.replace(" ","")
message = json.dumps({"mode": "online", "text": rec_result["text"], "wav_name": websocket.wav_name})
await websocket.send(message)

2
funasr/version.txt
View File

@ -1 +1 @@
0.4.5
0.4.6