From 8706e767affc6bdc8cb7a67ca3a20a62779ff048 Mon Sep 17 00:00:00 2001
From: 雾聪 <wucong.lyb@alibaba-inc.com>
Date: 星期三, 17 五月 2023 15:45:46 +0800
Subject: [PATCH] Merge branch 'main' of https://github.com/alibaba-damo-academy/FunASR into main
---
funasr/runtime/python/websocket/ws_server_online.py | 123 +++++++++++++++++++++++-----------------
1 files changed, 71 insertions(+), 52 deletions(-)
diff --git a/funasr/runtime/python/websocket/ws_server_online.py b/funasr/runtime/python/websocket/ws_server_online.py
index 6ea8f39..16a3abe 100644
--- a/funasr/runtime/python/websocket/ws_server_online.py
+++ b/funasr/runtime/python/websocket/ws_server_online.py
@@ -7,7 +7,7 @@
import logging
import tracemalloc
import numpy as np
-
+import ssl
from parse_args import args
from modelscope.pipelines import pipeline
from modelscope.utils.constant import Tasks
@@ -26,68 +26,87 @@
print("model loading")
inference_pipeline_asr_online = pipeline(
- task=Tasks.auto_speech_recognition,
- model=args.asr_model_online,
- ngpu=args.ngpu,
- ncpu=args.ncpu,
- model_revision='v1.0.4')
+ task=Tasks.auto_speech_recognition,
+ model=args.asr_model_online,
+ ngpu=args.ngpu,
+ ncpu=args.ncpu,
+ model_revision='v1.0.4')
print("model loaded")
async def ws_serve(websocket, path):
- frames_online = []
- global websocket_users
- websocket.send_msg = Queue()
- websocket_users.add(websocket)
- websocket.param_dict_asr_online = {"cache": dict()}
- websocket.speek_online = Queue()
-
- try:
- async for message in websocket:
- message = json.loads(message)
- is_finished = message["is_finished"]
- if not is_finished:
- audio = bytes(message['audio'], 'ISO-8859-1')
-
- is_speaking = message["is_speaking"]
- websocket.param_dict_asr_online["is_final"] = not is_speaking
-
- websocket.param_dict_asr_online["chunk_size"] = message["chunk_size"]
-
- frames_online.append(audio)
- if len(frames_online) % message["chunk_interval"] == 0 or not is_speaking:
- audio_in = b"".join(frames_online)
- await async_asr_online(websocket,audio_in)
- frames_online = []
+ frames_asr_online = []
+ global websocket_users
+ websocket_users.add(websocket)
+ websocket.param_dict_asr_online = {"cache": dict()}
+ websocket.wav_name = "microphone"
+ print("new user connected",flush=True)
+ try:
+ async for message in websocket:
+
+
+ if isinstance(message, str):
+ messagejson = json.loads(message)
+
+ if "is_speaking" in messagejson:
+ websocket.is_speaking = messagejson["is_speaking"]
+ websocket.param_dict_asr_online["is_final"] = not websocket.is_speaking
+ # need to fire engine manually if no data received any more
+ if not websocket.is_speaking:
+ await async_asr_online(websocket,b"")
+ if "chunk_interval" in messagejson:
+ websocket.chunk_interval=messagejson["chunk_interval"]
+ if "wav_name" in messagejson:
+ websocket.wav_name = messagejson.get("wav_name")
+ if "chunk_size" in messagejson:
+ websocket.param_dict_asr_online["chunk_size"] = messagejson["chunk_size"]
+ # if has bytes in buffer or message is bytes
+ if len(frames_asr_online) > 0 or not isinstance(message, str):
+ if not isinstance(message,str):
+ frames_asr_online.append(message)
+ if len(frames_asr_online) % websocket.chunk_interval == 0 or not websocket.is_speaking:
+ audio_in = b"".join(frames_asr_online)
+ # if not websocket.is_speaking:
+ #padding 0.5s at end gurantee that asr engine can fire out last word
+ # audio_in=audio_in+b''.join(np.zeros(int(16000*0.5),dtype=np.int16))
+ await async_asr_online(websocket,audio_in)
+ frames_asr_online = []
+
+
+ except websockets.ConnectionClosed:
+ print("ConnectionClosed...", websocket_users)
+ websocket_users.remove(websocket)
+ except websockets.InvalidState:
+ print("InvalidState...")
+ except Exception as e:
+ print("Exception:", e)
-
- except websockets.ConnectionClosed:
- print("ConnectionClosed...", websocket_users)
- websocket_users.remove(websocket)
- except websockets.InvalidState:
- print("InvalidState...")
- except Exception as e:
- print("Exception:", e)
-
async def async_asr_online(websocket,audio_in):
- if len(audio_in) > 0:
- audio_in = load_bytes(audio_in)
- rec_result = inference_pipeline_asr_online(audio_in=audio_in,
- param_dict=websocket.param_dict_asr_online)
- if websocket.param_dict_asr_online["is_final"]:
- 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"]})
- await websocket.send(message)
+ if len(audio_in) >=0:
+ audio_in = load_bytes(audio_in)
+ rec_result = inference_pipeline_asr_online(audio_in=audio_in,
+ param_dict=websocket.param_dict_asr_online)
+ if websocket.param_dict_asr_online.get("is_final", False):
+ 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":
+ message = json.dumps({"mode": "online", "text": rec_result["text"], "wav_name": websocket.wav_name})
+ await websocket.send(message)
+if len(args.certfile)>0:
+ ssl_context = ssl.SSLContext(ssl.PROTOCOL_TLS_SERVER)
-start_server = websockets.serve(ws_serve, args.host, args.port, subprotocols=["binary"], ping_interval=None)
+ # Generate with Lets Encrypt, copied to this location, chown to current user and 400 permissions
+ ssl_cert = args.certfile
+ ssl_key = args.keyfile
+
+ ssl_context.load_cert_chain(ssl_cert, keyfile=ssl_key)
+ start_server = websockets.serve(ws_serve, args.host, args.port, subprotocols=["binary"], ping_interval=None,ssl=ssl_context)
+else:
+ start_server = websockets.serve(ws_serve, args.host, args.port, subprotocols=["binary"], ping_interval=None)
asyncio.get_event_loop().run_until_complete(start_server)
asyncio.get_event_loop().run_forever()
\ No newline at end of file
--
Gitblit v1.9.1