pip install sounddevice soundfile -i https://pypi.tuna.tsinghua.edu.cn/simple

# 導(dǎo)入所需的庫
import sounddevice as sd
import soundfile as sf

# 定義測試麥克風(fēng)的函數(shù)
def test_microphone(device_index=None, output_filename="output.wav"):
    # 設(shè)置錄音參數(shù)
    duration = 5  # 錄音時長（秒）
    fs = 44100  # 采樣頻率

    # 獲取麥克風(fēng)列表
    devices = sd.query_devices()
    # 如果提供了設(shè)備索引并且有效，則使用指定的麥克風(fēng)
    if device_index is not None and device_index < len(devices):
        print(f"Using microphone: {devices[device_index]['name']}")
    else:
        print("Using default microphone.")

    # 獲取并設(shè)置麥克風(fēng)支持的采樣率
    supported_rates = devices[device_index]['default_samplerate']
    if supported_rates != fs:
        print(f"Adjusting sample rate to {int(supported_rates)} Hz (supported by the device).")
        fs = int(supported_rates)  # 確保采樣率是整數(shù)

    print("Recording...")
    # 使用sounddevice錄制聲音
    recording = sd.rec(int(duration * fs), samplerate=fs, channels=1, device=device_index)

    # 等待錄音完成
    sd.wait()

    print("Recording finished.")

    # 保存錄音為WAV文件
    sf.write(output_filename, recording, fs)

    print(f"File saved as {output_filename}")

# 主函數(shù)入口
if __name__ == "__main__":
    # 獲取麥克風(fēng)列表并打印
    devices = sd.query_devices()
    for i, device in enumerate(devices):
        print(f"{i}: {device['name']}")

    # 用戶輸入選擇麥克風(fēng)設(shè)備索引
    device_index = int(input("Enter the index of the microphone to use: "))
    test_microphone(device_index)

import time
import queue
import sounddevice as sd
import numpy as np
from aip import AipSpeech
import sys
 
# 百度云配置信息
APP_ID = '你的 App ID'  # 替換為實(shí)際的 APP ID
API_KEY = '你的 Api Key'  # 替換為實(shí)際的 API KEY
SECRET_KEY = '你的 Secret Key'  # 替換為實(shí)際的 SECRET KEY
 
client = AipSpeech(APP_ID, API_KEY, SECRET_KEY)
 
# Queue to hold the recorded audio data
audio_queue = queue.Queue()
speaker_queue = queue.Queue()
 
# Callback function to capture audio data from microphone
def audio_callback(indata, frames, time, status):
    if status:
        print(status, file=sys.stderr)
    audio_queue.put(indata.copy())
 
# Callback function to capture audio data from speaker
def speaker_callback(indata, frames, time, status):
    if status:
        print(status, file=sys.stderr)
    speaker_queue.put(indata.copy())

class RealTimeSpeechRecognizer:
    def __init__(self, client, name):
        self.client = client
        self.name = name
 
    def send_audio(self, audio_data):
        result = self.client.asr(audio_data, 'pcm', 16000, {
            'dev_pid': 1537,
        })
        if result.get('err_no') == 0:
            print(f"{self.name} 識別結(jié)果: {result['result']}")
        else:
            print(f"{self.name} 錯誤: {result['err_msg']}")
 
# 調(diào)用百度的語音轉(zhuǎn)文字的接口
def recognize_speech(audio_data, recognizer):
    audio_data = np.concatenate(audio_data)
    recognizer.send_audio(audio_data.tobytes())

def start_audio_stream(mic_recognizer, speaker_recognizer, speaker_device_index):
    with sd.InputStream(callback=audio_callback, channels=1, samplerate=16000, dtype='int16') as mic_stream, \
            sd.InputStream(callback=speaker_callback, channels=1, samplerate=16000, dtype='int16', device=speaker_device_index) as spk_stream:
        print("Recording audio... Press Ctrl+C to stop.")
        mic_audio_buffer = []
        speaker_audio_buffer = []
        try:
            while True:
                while not audio_queue.empty():
                    mic_audio_buffer.append(audio_queue.get())
                while not speaker_queue.empty():
                    speaker_audio_buffer.append(speaker_queue.get())
 
                if len(mic_audio_buffer) >= 10:
                    recognize_speech(mic_audio_buffer, mic_recognizer)
                    mic_audio_buffer = []  # Clear buffer after sending
 
                if len(speaker_audio_buffer) >= 10:
                    recognize_speech(speaker_audio_buffer, speaker_recognizer)
                    speaker_audio_buffer = []  # Clear buffer after sending
 
                time.sleep(0.1)
        except KeyboardInterrupt:
            print("Stopping audio recording.")

if __name__ == "__main__":
    speaker_device_index = 8  # 使用 pulse 設(shè)備（索引 8）來捕獲揚(yáng)聲器輸出
 
    mic_recognizer = RealTimeSpeechRecognizer(client, "麥克風(fēng)接收：")
    speaker_recognizer = RealTimeSpeechRecognizer(client, "揚(yáng)聲器接收：")
 
    start_audio_stream(mic_recognizer, speaker_recognizer, speaker_device_index)