From a30953f75997d31fbc1b18fedf4ec7f82b2f58cf Mon Sep 17 00:00:00 2001
From: zhaomingwork <zhaomingwork@qq.com>
Date: 星期五, 12 五月 2023 11:19:19 +0800
Subject: [PATCH] cpp python websocket compatible
---
funasr/runtime/python/websocket/ws_client.py | 19 +++++++++++++------
funasr/runtime/python/websocket/ws_server_online.py | 41 +++++++++++++++++++++++++++--------------
2 files changed, 40 insertions(+), 20 deletions(-)
diff --git a/funasr/runtime/python/websocket/ws_client.py b/funasr/runtime/python/websocket/ws_client.py
index a4a6d9f..f8632ce 100644
--- a/funasr/runtime/python/websocket/ws_client.py
+++ b/funasr/runtime/python/websocket/ws_client.py
@@ -84,12 +84,13 @@
rate=RATE,
input=True,
frames_per_buffer=CHUNK)
+ message = json.dumps({"chunk_size": args.chunk_size, "chunk_interval": args.chunk_interval, "wav_name": wav_name,"is_speaking": True})
+ voices.put(message)
is_speaking = True
while True:
data = stream.read(CHUNK)
- data = data.decode('ISO-8859-1')
- message = json.dumps({"chunk_size": args.chunk_size, "chunk_interval": args.chunk_interval, "audio": data, "is_speaking": is_speaking, "is_finished": is_finished})
+ message = data
voices.put(message)
@@ -122,15 +123,21 @@
stride = int(60*args.chunk_size[1]/args.chunk_interval/1000*16000*2)
chunk_num = (len(audio_bytes)-1)//stride + 1
# print(stride)
+
+ # send first time
+ message = json.dumps({"chunk_size": args.chunk_size, "chunk_interval": args.chunk_interval, "wav_name": wav_name,"is_speaking": True})
+ voices.put(message)
is_speaking = True
for i in range(chunk_num):
- if i == chunk_num-1:
- is_speaking = False
+
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, "wav_name": wav_name})
+ message = data
voices.put(message)
+ if i == chunk_num-1:
+ is_speaking = False
+ message = json.dumps({"is_speaking": is_speaking})
+ voices.put(message)
# print("data_chunk: ", len(data_chunk))
# print(voices.qsize())
sleep_duration = 0.001 if args.send_without_sleep else 60*args.chunk_size[1]/args.chunk_interval/1000
diff --git a/funasr/runtime/python/websocket/ws_server_online.py b/funasr/runtime/python/websocket/ws_server_online.py
index 3c0fb16..44edf98 100644
--- a/funasr/runtime/python/websocket/ws_server_online.py
+++ b/funasr/runtime/python/websocket/ws_server_online.py
@@ -41,25 +41,37 @@
global websocket_users
websocket_users.add(websocket)
websocket.param_dict_asr_online = {"cache": dict()}
-
+ print("new user connected",flush=True)
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.wav_name = message.get("wav_name", "demo")
- websocket.param_dict_asr_online["chunk_size"] = message["chunk_size"]
-
- frames_asr_online.append(audio)
- if len(frames_asr_online) % message["chunk_interval"] == 0 or not is_speaking:
+
+
+ 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
+ if "is_finished" in messagejson:
+ websocket.is_speaking = False
+ websocket.param_dict_asr_online["is_final"] = True
+ if "chunk_interval" in messagejson:
+ websocket.chunk_interval=messagejson["chunk_interval"]
+ if "wav_name" in messagejson:
+ websocket.wav_name = messagejson.get("wav_name", "demo")
+ 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:
@@ -69,6 +81,7 @@
print("InvalidState...")
except Exception as e:
print("Exception:", e)
+
async def async_asr_online(websocket,audio_in):
if len(audio_in) > 0:
--
Gitblit v1.9.1