实时语音识别技术实现

实时语音识别

1.环境
2.完整代码
3.效果
4.可能的问题

实时语音识别

1.环境

python版本：3.11.9

2.完整代码

import sqlite3
import timefrom funasr import AutoModel
import sounddevice as sd
import numpy as np# 模型参数设置
chunk_size = [0, 10, 5]
encoder_chunk_look_back = 7
decoder_chunk_look_back = 5model = AutoModel(model="D:\SpeechRecognize\speech_paraformer-large-vad-punc_asr_nat-zh-cn-16k-common-vocab8404-pytorch")# 假设模型要求的采样率为 16000
fs = 16000
duration = 3 #时间
chunk_stride = chunk_size[1] * 960
cache = {}
window_size = 3# 连接到 SQLite 数据库，如果不存在则会创建新的数据库文件
conn = sqlite3.connect('speech_recognition.db')
cursor = conn.cursor()# 创建表格
cursor.execute('''CREATE TABLE IF NOT EXISTS speech_data(text TEXT, time_stamp TEXT, batch TEXT)
''')while True:start_time = time.strftime("%Y-%m-%d %H:%M:%S", time.localtime())myrecording = sd.rec(int(fs * duration), samplerate=fs, channels=1)sd.wait()speech_chunk = myrecording.flatten()# 噪声处理filtered_chunk = np.convolve(speech_chunk, np.ones(window_size) / window_size, mode='same')speech_chunk = filtered_chunkis_final = Falseres = model.generate(input=speech_chunk, cache=cache, is_final=is_final, chunk_size=chunk_size,encoder_chunk_look_back=encoder_chunk_look_back,decoder_chunk_look_back=decoder_chunk_look_back)text_result=res[0]['text']print(text_result)cursor.execute("INSERT INTO speech_data (text, time_stamp, batch) VALUES (?,?,?)",(text_result, start_time, 'eerr'))conn.commit()