Python使用SpeechRecognition库实现语音识别功能

# 1、核心库安装
pip install SpeechRecognition

# 2、该库处理麦克风实时语音需要依赖`PyAudio`：
pip install pyaudio
# 若安装失败（Windows 常见）：先下载对应 Python 版本的PyAudio whl 文件（Unofficial Windows Binaries），再本地安装
pip install PyAudio-0.2.13-cp310-cp310-win_amd64.whl

# 3、Mac 用户：先装底层依赖再装 PyAudio
brew install portaudio
pip install pyaudio

# 4、若需离线识别，额外安装
pip install pocketsphinx

import speech_recognition as sr

def microphone_recognition():
    # 1. 创建Recognizer实例
    r = sr.Recognizer()
    
    # 2. 获取麦克风设备（默认麦克风）
    with sr.Microphone() as source:
        print("正在校准环境噪声，请保持安静...")
        # 校准噪声：监听1秒，获取背景噪声水平（关键步骤，否则识别准确率低）
        r.adjust_for_ambient_noise(source, duration=1)
        print("校准完成，请开始说话（说完后会自动识别）...")
        
        # 3. 监听麦克风音频（timeout=5表示5秒内无语音则超时）
        audio = r.listen(source, timeout=5, phrase_time_limit=10)  # phrase_time_limit限制单次语音最长10秒
    
    # 4. 调用Google API识别语音
    try:
        print("正在识别...")
        # language="zh-CN"指定中文识别，默认是英文
        # 若识别英文可改为`language="en-US"`。
        text = r.recognize_google(audio, language="zh-CN")
        print(f"你说的内容：{text}")
    except sr.WaitTimeoutError:
        print("错误：超过5秒未检测到语音输入")
    except sr.UnknownValueError:
        print("错误：无法识别你说的内容（语音模糊/无有效语音）")
    except sr.RequestError as e:
        print(f"错误：调用Google API失败 → {e}")

if __name__ == "__main__":
    microphone_recognition()

import speech_recognition as sr

def audio_file_recognition(file_path):
    # 1. 创建Recognizer实例
    r = sr.Recognizer()
    
    # 2. 读取音频文件
    with sr.AudioFile(file_path) as source:
        # 读取完整音频（也可指定时长：r.record(source, duration=5) 读取前5秒）
        audio = r.record(source)
    
    # 3. 识别音频
    try:
        text = r.recognize_google(audio, language="zh-CN")
        print(f"音频内容：{text}")
    except sr.UnknownValueError:
        print("错误：无法识别音频内容")
    except sr.RequestError as e:
        print(f"错误：API调用失败 → {e}")

if __name__ == "__main__":
    # 替换为你的音频文件路径（建议用wav格式）
    audio_file_recognition("test_audio.wav")

pip install pydub
# Windows需下载ffmpeg并配置环境变量，Mac：brew install ffmpeg

import speech_recognition as sr
from pydub import AudioSegment

# 转换MP3为WAV
audio = AudioSegment.from_mp3("test_audio.mp3")
audio.export("test_audio_wav.wav", format="wav")

# 调用上面的识别函数
audio_file_recognition("test_audio_wav.wav")

import speech_recognition as sr

def offline_recognition():
    r = sr.Recognizer()
    with sr.Microphone() as source:
        r.adjust_for_ambient_noise(source, duration=1)
        print("离线模式，请说话...")
        audio = r.listen(source)
    
    try:
        # 离线识别（中文需额外配置语言包，默认英文）
        text = r.recognize_sphinx(audio, language="zh-CN")
        print(f"离线识别结果：{text}")
    except sr.UnknownValueError:
        print("错误：离线识别失败（语音无法解析）")
    except sr.RequestError as e:
        print(f"错误：Sphinx引擎初始化失败 → {e}")

if __name__ == "__main__":
    offline_recognition()

   pip install baidu-aip

import speech_recognition as sr
from aip import AipSpeech

# 替换为你的百度应用信息
APP_ID = "你的APP_ID"
API_KEY = "你的API_KEY"
SECRET_KEY = "你的SECRET_KEY"

def baidu_recognition(audio_data):
    # 初始化百度语音客户端
    client = AipSpeech(APP_ID, API_KEY, SECRET_KEY)
    
    # 将AudioData转为百度要求的格式（pcm/16k/16位/单声道）
    pcm_data = audio_data.get_raw_data(convert_rate=16000, convert_width=2)
    
    # 调用百度API识别
    result = client.asr(pcm_data, "pcm", 16000, {
        "dev_pid": 1536,  # 1536=普通话（支持简单的英文），1537=英语
    })
    
    if result["err_no"] == 0:
        return "".join(result["result"])
    else:
        raise Exception(f"百度API错误：{result['err_msg']}")

# 主流程：麦克风采集+百度识别
if __name__ == "__main__":
    r = sr.Recognizer()
    with sr.Microphone(sample_rate=16000) as source:  # 百度要求16k采样率
        r.adjust_for_ambient_noise(source, duration=1)
        print("请说话（百度语音识别）...")
        audio = r.listen(source)
    
    try:
        text = baidu_recognition(audio)
        print(f"百度识别结果：{text}")
    except Exception as e:
        print(f"识别失败：{e}")