ai vosk whisper pyaudio fastapi 【小沐学Python】Python实现语音识别（SpeechRecognition）

文章目录

1、简介2、安装和测试2.1 安装python2.2 安装SpeechRecognition2.3 安装pyaudio2.4 安装pocketsphinx（offline）2.5 安装Vosk （offline）2.6 安装Whisper（offline）

3 测试3.1 命令3.2 fastapi3.3 google3.4 recognize_sphinx3.5 语音生成音频文件

结语

1、简介

https://pypi.org/project/SpeechRecognition/ https://github.com/Uberi/speech_recognition

SpeechRecognition用于执行语音识别的库，支持多个引擎和 API，在线和离线。

Speech recognition engine/API 支持如下接口:

recognize_bing()：Microsoft Bing Speech

recognize_google()： Google Web Speech API

recognize_google_cloud()：Google Cloud Speech - requires installation of the google-cloud-speech package

recognize_houndify()： Houndify by SoundHound

recognize_ibm()：IBM Speech to Text

recognize_sphinx()：CMU Sphinx - requires installing PocketSphinx

recognize_wit()：Wit.ai

以上几个中只有 recognition_sphinx（）可与CMU Sphinx 引擎脱机工作， 其他六个都需要连接互联网。另外，SpeechRecognition 附带 Google Web Speech API 的默认 API 密钥，可直接使用它。其他的 API 都需要使用 API 密钥或用户名/密码组合进行身份验证。

2、安装和测试

Python 3.8+ (required) PyAudio 0.2.11+ (required only if you need to use microphone input, Microphone) PocketSphinx (required only if you need to use the Sphinx recognizer, recognizer_instance.recognize_sphinx) Google API Client Library for Python (required only if you need to use the Google Cloud Speech API, recognizer_instance.recognize_google_cloud) FLAC encoder (required only if the system is not x86-based Windows/Linux/OS X) Vosk (required only if you need to use Vosk API speech recognition recognizer_instance.recognize_vosk) Whisper (required only if you need to use Whisper recognizer_instance.recognize_whisper) openai (required only if you need to use Whisper API speech recognition recognizer_instance.recognize_whisper_api)

2.1 安装python

https://www.python.org/downloads/

2.2 安装SpeechRecognition

安装库SpeechRecognition：

#python -m pip install --upgrade pip

#pip install 包名 -i https://pypi.tuna.tsinghua.edu.cn/simple/

#pip install 包名 -i http://pypi.douban.com/simple/ --trusted-host pypi.douban.com

#pip install 包名 -i https://pypi.org/simple

pip install SpeechRecognition

import speech_recognition as sr

print(sr.__version__)

麦克风的特定于硬件的索引获取：

import speech_recognition as sr

for index, name in enumerate(sr.Microphone.list_microphone_names()):

print("Microphone with name \"{1}\" found for `Microphone(device_index={0})`".format(index, name))

2.3 安装pyaudio

pip install pyaudio

2.4 安装pocketsphinx（offline）

pip install pocketsphinx

或者https://www.lfd.uci.edu/~gohlke/pythonlibs/#pocketsphinx找到编译好的本地库文件进行安装。 在这里使用的是recognize_sphinx()语音识别器，它可以脱机工作，但是必须安装pocketsphinx库. 若要进行中文识别，还需要两样东西。 1、语音文件（SpeechRecognition对文件格式有要求）； SpeechRecognition支持语音文件类型：

WAV: 必须是 PCM/LPCM 格式

AIFF

AIFF-C

FLAC: 必须是初始 FLAC 格式；OGG-FLAC 格式不可用

2、中文声学模型、语言模型和字典文件; pocketsphinx需要安装的中文语言、声学模型。

https://sourceforge.net/projects/cmusphinx/files/Acoustic%20and%20Language%20Models/Mandarin/

下载cmusphinx-zh-cn-5.2.tar.gz并解压： 在python安装目录下找到Lib\site-packages\speech_recognition：

点击进入pocketsphinx-data文件夹，并新建文件夹zh-CN： 在这个文件夹中添加进入刚刚解压的文件，需要注意：把解压出来的zh_cn.cd_cont_5000文件夹重命名为acoustic-model、zh_cn.lm.bin命名为language-model.lm.bin、zh_cn.dic中dic改为pronounciation-dictionary.dict格式。

编写脚本测试：

import speech_recognition as sr

r = sr.Recognizer() #调用识别器

test = sr.AudioFile("chinese.flac") #导入语音文件

with test as source:

# r.adjust_for_ambient_noise(source)

audio = r.record(source) #使用 record() 从文件中获取数据

type(audio)

# c=r.recognize_sphinx(audio, language='zh-cn') #识别输出

c=r.recognize_sphinx(audio, language='en-US') #识别输出

print(c)

import speech_recognition as sr

# obtain path to "english.wav" in the same folder as this script

from os import path

AUDIO_FILE = path.join(path.dirname(path.realpath(__file__)), "english.wav")

# AUDIO_FILE = path.join(path.dirname(path.realpath(__file__)), "french.aiff")

# AUDIO_FILE = path.join(path.dirname(path.realpath(__file__)), "chinese.flac")

# use the audio file as the audio source

r = sr.Recognizer()

with sr.AudioFile(AUDIO_FILE) as source:

audio = r.record(source) # read the entire audio file

# recognize speech using Sphinx

try:

print("Sphinx thinks you said " + r.recognize_sphinx(audio))

except sr.UnknownValueError:

print("Sphinx could not understand audio")

except sr.RequestError as e:

print("Sphinx error; {0}".format(e))

import speech_recognition as sr

recognizer = sr.Recognizer()

with sr.Microphone() as source:

# recognizer.adjust_for_ambient_noise(source)

audio = recognizer.listen(source)

c=recognizer.recognize_sphinx(audio, language='zh-cn') #识别输出

# c=r.recognize_sphinx(audio, language='en-US') #识别输出

print(c)

import speech_recognition as sr

# obtain audio from the microphone

r = sr.Recognizer()

with sr.Microphone() as source:

print("Say something!")

audio = r.listen(source)

# recognize speech using Sphinx

try:

print("Sphinx thinks you said " + r.recognize_sphinx(audio))

except sr.UnknownValueError:

print("Sphinx could not understand audio")

except sr.RequestError as e:

print("Sphinx error; {0}".format(e))

2.5 安装Vosk （offline）

python3 -m pip install vosk

您还必须安装 Vosk 模型： 以下是可供下载的模型。您必须将它们放在项目的模型文件夹中，例如“your-project-folder/models/your-vosk-model” https://alphacephei.com/vosk/models

在测试脚本的所在文件夹，新建model子文件夹，然后把上面下载的模型解压到里面如下： 编写脚本：

import speech_recognition as sr

from vosk import KaldiRecognizer, Model

r = sr.Recognizer()

with sr.Microphone() as source:

audio = r.listen(source, timeout=3, phrase_time_limit=3)

r.vosk_model = Model(model_name="vosk-model-small-cn-0.22")

text=r.recognize_vosk(audio, language='zh-cn')

print(text)

2.6 安装Whisper（offline）

pip install zhconv

pip install whisper

pip install -U openai-whisper

pip3 install wheel

pip install soundfile

编写脚本：

import speech_recognition as sr

from vosk import KaldiRecognizer, Model

r = sr.Recognizer()

with sr.Microphone() as source:

audio = r.listen(source, timeout=3, phrase_time_limit=5)

# recognize speech using whisper

try:

print("Whisper thinks you said: " + r.recognize_whisper(audio, language="chinese"))

except sr.UnknownValueError:

print("Whisper could not understand audio")

except sr.RequestError as e:

print("Could not request results from Whisper")

3 测试

3.1 命令

python -m speech_recognition

3.2 fastapi

import json

import os

from pprint import pprint

import speech_recognition

import torch

import uvicorn

from fastapi import FastAPI, HTTPException

from pydantic import BaseModel

import soundfile

import whisper

import vosk

class ResponseModel(BaseModel):

path: str

app = FastAPI()

def get_path(req: ResponseModel):

path = req.path

if path == "":

raise HTTPException(status_code=400, detail="No path provided")

if not path.endswith(".wav"):

raise HTTPException(status_code=400, detail="Invalid file type")

if not os.path.exists(path):

raise HTTPException(status_code=404, detail="File does not exist")

return path

@app.get("/")

def root():

return {"message": "speech-recognition api"}

@app.post("/recognize-google")

def recognize_google(req: ResponseModel):

path = get_path(req)

r = speech_recognition.Recognizer()

with speech_recognition.AudioFile(path) as source:

audio = r.record(source)

return r.recognize_google(audio, language='ja-JP', show_all=True)

@app.post("/recognize-vosk")

def recognize_vosk(req: ResponseModel):

path = get_path(req)

r = speech_recognition.Recognizer()

with speech_recognition.AudioFile(path) as source:

audio = r.record(source)

return json.loads(r.recognize_vosk(audio, language='ja'))

@app.post("/recognize-whisper")

def recognize_whisper(req: ResponseModel):

path = get_path(req)

r = speech_recognition.Recognizer()

with speech_recognition.AudioFile(path) as source:

audio = r.record(source)

result = r.recognize_whisper(audio, language='ja')

try:

return json.loads(result)

except:

return {"text": result}

if __name__ == "__main__":

host = os.environ.get('HOST', '0.0.0.0')

port: int = os.environ.get('PORT', 8080)

uvicorn.run("main:app", host=host, port=int(port))

3.3 google

import speech_recognition as sr

import webbrowser as wb

import speak

chrome_path = 'C:/Program Files (x86)/Google/Chrome/Application/chrome.exe %s'

r = sr.Recognizer()

with sr.Microphone() as source:

print ('Say Something!')

audio = r.listen(source)

print ('Done!')

try:

text = r.recognize_google(audio)

print('Google thinks you said:\n' + text)

lang = 'en'

speak.tts(text, lang)

f_text = 'https://www.google.co.in/search?q=' + text

wb.get(chrome_path).open(f_text)

except Exception as e:

print (e)

3.4 recognize_sphinx

import logging

import speech_recognition as sr

def audio_Sphinx(filename):

logging.info('开始识别语音文件...')

# use the audio file as the audio source

r = sr.Recognizer()

with sr.AudioFile(filename) as source:

audio = r.record(source) # read the entire audio file

# recognize speech using Sphinx

try:

print("Sphinx thinks you said: " + r.recognize_sphinx(audio, language='zh-cn'))

except sr.UnknownValueError:

print("Sphinx could not understand audio")

except sr.RequestError as e:

print("Sphinx error; {0}".format(e))

if __name__ == "__main__":

logging.basicConfig(level=logging.INFO)

wav_num = 0

while True:

r = sr.Recognizer()

#启用麦克风

mic = sr.Microphone()

logging.info('录音中...')

with mic as source:

#降噪

r.adjust_for_ambient_noise(source)

audio = r.listen(source)

with open(f"00{wav_num}.wav", "wb") as f:

#将麦克风录到的声音保存为wav文件

f.write(audio.get_wav_data(convert_rate=16000))

logging.info('录音结束，识别中...')

target = audio_Sphinx(f"00{wav_num}.wav")

wav_num += 1

3.5 语音生成音频文件

方法1：

import speech_recognition as sr

# Use SpeechRecognition to record 使用语音识别包录制音频

def my_record(rate=16000):

r = sr.Recognizer()

with sr.Microphone(sample_rate=rate) as source:

print("please say something")

audio = r.listen(source)

with open("voices/myvoices.wav", "wb") as f:

f.write(audio.get_wav_data())

print("录音完成！")

my_record()

方法2：

import wave

from pyaudio import PyAudio, paInt16

framerate = 16000 # 采样率

num_samples = 2000 # 采样点

channels = 1 # 声道

sampwidth = 2 # 采样宽度2bytes

FILEPATH = 'voices/myvoices.wav'

def save_wave_file(filepath, data):

wf = wave.open(filepath, 'wb')

wf.setnchannels(channels)

wf.setsampwidth(sampwidth)

wf.setframerate(framerate)

wf.writeframes(b''.join(data))

wf.close()

#录音

def my_record():

pa = PyAudio()

#打开一个新的音频stream

stream = pa.open(format=paInt16, channels=channels,

rate=framerate, input=True, frames_per_buffer=num_samples)

my_buf = [] #存放录音数据

t = time.time()

print('正在录音...')

while time.time() < t + 10: # 设置录音时间（秒）

#循环read，每次read 2000frames

string_audio_data = stream.read(num_samples)

my_buf.append(string_audio_data)

print('录音结束.')

save_wave_file(FILEPATH, my_buf)

stream.close()

结语

如果您觉得该方法或代码有一点点用处，可以给作者点个赞，或打赏杯咖啡；╮(￣▽￣)╭ 如果您感觉方法或代码不咋地//(ㄒoㄒ)//，就在评论处留言，作者继续改进；o_O??? 如果您需要相关功能的代码定制化开发，可以留言私信作者；(✿◡‿◡) 感谢各位大佬童鞋们的支持！( ´ ▽´ )ﾉ ( ´ ▽´)っ！！！

相关链接

 您阅读本篇文章共花了：