From 378ced7edb0cb9957263675215192e19e490d226 Mon Sep 17 00:00:00 2001
From: dyyzhmm <dyyzhmm@163.com>
Date: 星期三, 15 三月 2023 10:10:57 +0800
Subject: [PATCH] Merge pull request #1 from alibaba-damo-academy/dev_hmm
---
funasr/runtime/python/grpc/grpc_main_client_mic.py | 31 ++++++++++---------------------
1 files changed, 10 insertions(+), 21 deletions(-)
diff --git a/funasr/runtime/python/grpc/grpc_main_client_mic.py b/funasr/runtime/python/grpc/grpc_main_client_mic.py
index acbe90b..220e8b5 100644
--- a/funasr/runtime/python/grpc/grpc_main_client_mic.py
+++ b/funasr/runtime/python/grpc/grpc_main_client_mic.py
@@ -1,7 +1,6 @@
import pyaudio
import grpc
import json
-import webrtcvad
import time
import asyncio
import argparse
@@ -11,24 +10,16 @@
async def deal_chunk(sig_mic):
global stub,SPEAKING,asr_user,language,sample_rate
- if vad.is_speech(sig_mic, sample_rate): #speaking
- SPEAKING = True
- response = transcribe_audio_bytes(stub, sig_mic, user=asr_user, language=language, speaking = True, isEnd = False) #speaking, send audio to server.
- else: #silence
- begin_time = 0
- if SPEAKING: #means we have some audio recorded, send recognize order to server.
- SPEAKING = False
- begin_time = int(round(time.time() * 1000))
- response = transcribe_audio_bytes(stub, None, user=asr_user, language=language, speaking = False, isEnd = False) #speak end, call server for recognize one sentence
- resp = response.next()
- if "decoding" == resp.action:
- resp = response.next() #TODO, blocking operation may leads to miss some audio clips. C++ multi-threading is preferred.
- if "finish" == resp.action:
- end_time = int(round(time.time() * 1000))
- print (json.loads(resp.sentence))
- print ("delay in ms: %d " % (end_time - begin_time))
- else:
- pass
+
+ SPEAKING = True
+ resp = transcribe_audio_bytes(stub, sig_mic, user=asr_user, language=language, speaking = True, isEnd = False) #speaking, send audio to server.
+
+ if "decoding" == resp.action: #partial result
+ print(json.loads(resp.sentence))
+ elif "finish" == resp.action: #final result
+ print (json.loads(resp.sentence))
+
+
async def record(host,port,sample_rate,mic_chunk,record_seconds,asr_user,language):
@@ -88,8 +79,6 @@
language = 'zh-CN'
- vad = webrtcvad.Vad()
- vad.set_mode(1)
FORMAT = pyaudio.paInt16
CHANNELS = 1
--
Gitblit v1.9.1