目录

1. 简介

2. 安装

2.1 安装speechRecognition

2.2 安装playsound

2.3 安装sounddevice

2.4 安装PocketSphinx

3. 驱动麦克风(录音)

4. 语音识别示例

4.1 英文识别

4.2 中文识别

4.3 去噪API

1. 简介

SpeechRecognition模块里面有接口可以实现麦克风数据读取、音频数据转换、语音识别等接口。

接口说明见连接：https://github.com/Uberi/speech_recognition#readme

2. 安装

2.1 安装speechRecognition

可以选择在conda环境下去安装，命令为：pip install speechRecognition -i 国内镜像源

国内镜像源：

清华：https://pypi.tuna.tsinghua.edu.cn/simple/

中科大：https://pypi.mirrors.ustc.edu.cn/simple/

豆瓣：https://pypi.doubanio.com/simple/

安装完毕

验证版本：

(cpu_env) C:\Users\Administrator>python

Python 3.6.5 | packaged by conda-forge | (default, Apr  6 2018, 16:13:55) [MSC v.1900 64 bit (AMD64)] on win32

Type "help", "copyright", "credits" or "license" for more information.

>>> import speech_recognition as asr

>>> asr.__version__

'3.9.0'

>>> exit()

(cpu_env) C:\Users\Administrator>

2.2 安装playsound

(cpu_env) C:\Users\Administrator>pip install playsound -i https://pypi.doubanio.com/simple/

Looking in indexes: https://pypi.doubanio.com/simple/

Collecting playsound

  Downloading https://mirrors.cloud.tencent.com/pypi/packages/67/8c/c9f46b4b194126c4abb12e96321a6bea5c8dcc5c0e4d97622c14dfabe299/playsound-1.3.0.tar.gz (7.7 kB)

  Preparing metadata (setup.py) ... done

Building wheels for collected packages: playsound

  Building wheel for playsound (setup.py) ... done

  Created wheel for playsound: filename=playsound-1.3.0-py3-none-any.whl size=7035 sha256=f910d39a3d4354380699118a34100b211fc71377aa8d57d2a1bd025a85d7233c

  Stored in directory: c:\users\administrator\appdata\local\pip\cache\wheels\57\52\c1\4570be9fbeb4e664370f6f09428572bcf5c1afa63749bb1da4

Successfully built playsound

Installing collected packages: playsound

Successfully installed playsound-1.3.0

(cpu_env) C:\Users\Administrator>

2.3 安装sounddevice

sounddevice可以代替pyaudio，并且sounddevice的参数更加丰富。

(cpu_env) C:\Users\Administrator>pip install sounddevice -i https://pypi.mirrors.ustc.edu.cn/simple/

Looking in indexes: https://pypi.mirrors.ustc.edu.cn/simple/

Collecting sounddevice

  Downloading https://mirrors.ustc.edu.cn/pypi/packages/c0/36/1f284700e79d212996b80448de8e7c13a662ad3f910dec58ba8e249ff67b/sounddevice-0.4.4-py3-none-win_amd64.whl (195 kB)

     |████████████████████████████████| 195 kB 384 kB/s

Requirement already satisfied: CFFI>=1.0 in c:\users\administrator\miniconda3\envs\cpu_env\lib\site-packages (from sounddevice) (1.15.1)

Requirement already satisfied: pycparser in c:\users\administrator\miniconda3\envs\cpu_env\lib\site-packages (from CFFI>=1.0->sounddevice) (2.21)

Installing collected packages: sounddevice

Successfully installed sounddevice-0.4.4

(cpu_env) C:\Users\Administrator>

2.4 安装PocketSphinx

(cpu_env) G:\book_gitcode\book\ch5>pip install pocketsphinx -i https://pypi.mirrors.ustc.edu.cn/simple/

Looking in indexes: https://pypi.mirrors.ustc.edu.cn/simple/

Collecting pocketsphinx

WARNING: Discarding https://mirrors.ustc.edu.cn/pypi/packages/db/d8/8147df2687a14eb87bfc0e8c2eb921bc4c29d2c6568e1bb08fdd2e2d1471/pocketsphinx-5.0.3.tar.gz#sha256=27f4de0ca2d2bce391ce87eaab84fe6f0bc059b306fd1702d5fe6549b66e1586 (from https://mirrors.ustc.edu.cn/pypi/simple/pocketsphinx/). Command errored out with exit status 1: 'C:\Users\Administrator\miniconda3\envs\cpu_env\python.exe' 'C:\Users\Administrator\miniconda3\envs\cpu_env\lib\site-packages\pip' install --ignore-installed --no-user --prefix 'C:\Users\ADMINI~1\AppData\Local\Temp\pip-build-env-qa5fnqvi\overlay' --no-warn-script-location --no-binary :none: --only-binary :none: -i https://pypi.mirrors.ustc.edu.cn/simple/ -- scikit-build-core Cython Check the logs for full command output.

  Downloading https://mirrors.ustc.edu.cn/pypi/packages/e0/6d/a46f9500b38f2844b15d861a2b5a6bea248b3026fafc8eb94b56d60cffc5/pocketsphinx-5.0.2.tar.gz (34.2 MB)

     |████████████████████████████████| 34.2 MB 6.4 MB/s

  Installing build dependencies ... done

  Getting requirements to build wheel ... done

  Preparing metadata (pyproject.toml) ... done

Requirement already satisfied: sounddevice in c:\users\administrator\miniconda3\envs\cpu_env\lib\site-packages (from pocketsphinx) (0.4.4)

Requirement already satisfied: CFFI>=1.0 in c:\users\administrator\miniconda3\envs\cpu_env\lib\site-packages (from sounddevice->pocketsphinx) (1.15.1)

Requirement already satisfied: pycparser in c:\users\administrator\miniconda3\envs\cpu_env\lib\site-packages (from CFFI>=1.0->sounddevice->pocketsphinx) (2.21)

Building wheels for collected packages: pocketsphinx

  Building wheel for pocketsphinx (pyproject.toml) ... done

  Created wheel for pocketsphinx: filename=pocketsphinx-5.0.2-cp36-cp36m-win_amd64.whl size=29048875 sha256=63888d5b17ba708da577c9d62a846d5e08a77bd8a4813d661c4aebe980ee849b

  Stored in directory: c:\users\administrator\appdata\local\pip\cache\wheels\40\05\9d\7f252b34648da667e1d3eaad1428ddaebda2dfc30650ee6493

Successfully built pocketsphinx

Installing collected packages: pocketsphinx

Successfully installed pocketsphinx-5.0.2

(cpu_env) G:\book_gitcode\book\ch5>

3. 驱动麦克风(录音)

示例代码：

import sounddevice as sd
import numpy as np
from scipy.io.wavfile import write

fs = 16000  # 16kHz采样率
duration = 5  # 录制5秒

print("开始录音...")
recording = sd.rec(int(duration * fs),
    samplerate=fs,
    channels=1,          # 单声道
    dtype='int16',       # 16bit
    blocking=True        
)

print("录音结束")
# 播放录音（保持相同参数）
sd.play(recording, fs, blocking=True)
write("output.wav", fs, recording)

4. 语音识别示例

Sphinx公司提供的sphinx接口不需要联网也能识别语音，可以是关键词，也可以是整个长句。

4.1 英文识别

示例代码：

import speech_recognition as sr

#创建对象
r=sr.Recognizer()
#读音频数据(单个关键词)
with sr.AudioFile('audio_file/backward.wav') as audio:
    audio_data=r.record(audio) #读音频数据

#识别音频
result_txt=r.recognize_sphinx(audio_data)
print(r'result=', result_txt)

#读音频数据(一个英文长句)
with sr.AudioFile('audio_file/harvard.wav') as audio:
    audio_data=r.record(audio) #读音频数据

#识别音频
result_txt=r.recognize_sphinx(audio_data)
print(r'result=', result_txt)

结果：

(cpu_env) G:\book_gitcode\book\ch5>python speech_recog.py

result= backward

result= this they'll smell of old we're lingers it takes heat to bring out the odor they called it restores health and zest case all the gold he's fine with him tacos all pastore my favorite he's as full food is the hot cross bonn

(cpu_env) G:\book_gitcode\book\ch5>

4.2 中文识别

默认安装的是英文识别模型，中文识别模型还需要手动下载模型文件。

访问：

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

单击下载文件

下载完毕：

将解压后的文件夹重命名为zh-CN，然后将整个文件复制到该目录下(以Miniconda开发环境为例)：

C:\Users\Administrator\miniconda3\envs\cpu_env\Lib\site-packages\speech_recognition\pocketsphinx-data

重命名文件夹和文件：

zh_cn.cd_cont_5000改为acoustic-model，这个目录下是声学模型文件

zh_cn.dic改为pronounciation-dictionary.dict，这是发音词典文件

zh_cn.lm.bin改为language-model.lm.bin，这是语音模型文件

测试代码：

import speech_recognition as sr
#创建对象
r=sr.Recognizer()
#读音频数据(中文)
with sr.AudioFile('audio_file/nihao_wenwen_male.wav') as audio:
    audio_data=r.record(audio) #读音频数据

#识别音频
result_txt=r.recognize_sphinx(audio_data, language='zh-CN')
print(r'result=', result_txt)

4.3 去噪API

接口adjust_for_ambient_noise()可以对输入语音进行增强，先去噪再做识别。

示例代码：同

# python # 语音识别
Logo
腾讯云开发者社区

腾讯云面向开发者汇聚海量精品云计算使用和开发经验，营造开放的云计算技术生态圈。

更多推荐

Elasticsearch复杂数据类型终极指南：从入门到精通

Elasticsearch作为功能强大的搜索引擎，支持多种复杂数据类型，让开发者能够灵活处理各种结构化和非结构化数据。本文将带你全面了解Elasticsearch中的复杂数据类型，从基础概念到实际应用，助你轻松掌握数据建模的核心技巧。## 内部对象：构建层级化数据结构在Elasticsearch中，对象类型（Object）是最基础的复杂数据类型之一，用于表示具有嵌套关系的数据。例如，我们可

avatar 腾讯云开发者社区

终极指南：Flink SQL连接器版本管理从混乱到有序的升级之路

Apache Flink作为流处理领域的佼佼者，其SQL连接器的版本管理一直是开发者面临的核心挑战。本文将系统讲解Flink SQL连接器版本管理的最佳实践，帮助你轻松应对版本兼容性问题，实现从混乱到有序的升级之旅。## 连接器版本管理的常见痛点 😫在Flink应用开发中，连接器版本管理常常让开发者头疼不已。不同版本的连接器可能导致各种兼容性问题，例如API变更、功能差异甚至运行时错误。

avatar 腾讯云开发者社区

如何快速搭建Neon无服务器PostgreSQL：面向初学者的完整指南

Neon是一款革命性的无服务器PostgreSQL解决方案，它通过分离存储和计算层，实现了自动扩缩容、类代码式数据库分支以及零级扩展能力。本指南将帮助你从零开始搭建Neon开发环境，体验这款创新数据库的强大功能。## 准备工作：环境要求与依赖项在开始搭建Neon环境前，请确保你的系统满足以下要求：- Linux操作系统（推荐Ubuntu 20.04+或Debian 11+）- Git

avatar 腾讯云开发者社区