本地部署语音转文字（whisper，SpeechRecognition）

本地部署语音转文字

1.whisper

1.首先安装Chocolatey

https://github.com/openai/whisper
以管理员身份运行PowerShell

Set-ExecutionPolicy Bypass -Scope Process -Force; [System.Net.ServicePointManager]::SecurityProtocol = [System.Net.ServicePointManager]::SecurityProtocol -bor 3072; iex ((New-Object System.Net.WebClient).DownloadString('https://chocolatey.org/install.ps1'))
choco -v#查看版本确定安装成功
Set-ExecutionPolicy AllSigned
Set-ExecutionPolicy Bypass -Scope Process -Force; [System.Net.ServicePointManager]::SecurityProtocol = [System.Net.ServicePointManager]::SecurityProtocol -bor 3072; iex ((New-Object System.Net.WebClient).DownloadString('https://community.chocolatey.org/install.ps1'))

2.安装

choco install ffmpeg
pip install setuptools-rust
pip install -U openai-whisper

3.使用

whisper 1207.m4a --language Chinese --model medium --device cuda:0 

这时会自动下载模型文件到用户的.cache文件夹里面，我的电脑在C:\Users\robot\.cache\whisper。

2.SpeechRecognition

1.环境

pip install SpeechRecognition
pip install -i https://pypi.tuna.tsinghua.edu.cn/simple pocketsphinx

2.中文包

在这个虚拟环境的\site-packages\speech_recognition\pocketsphinx-data目录下添加zh-CN中文包。

3.格式转化

因为这个离线只支持一下三种格式，所以利用ffmpeg将语音转换成wav格式，这个包在装whisper的时候已经装好了，所以直接用：

 ffmpeg -i 12072.m4a output.wav

4.运行

import speech_recognition as sr
audio_file = 'output.wav'
r = sr.Recognizer()
with sr.AudioFile(audio_file) as source:
    audio = r.record(source)
# 识别音频文件
result = r.recognize_sphinx(audio, language="zh-CN")
print(result)

3.效果