audio_codebuddy/hardware-integration/pc-simulator/asr.py

# ASR 语音识别模块

from openai import OpenAI
from config import Config


def transcribe(audio_data: bytes) -> str:
    """
    把音频转为文字

    输入：WAV 格式音频 bytes
    输出：识别出的文字
    """
    if Config.ASR_PROVIDER == 'openai_whisper':
        return transcribe_whisper(audio_data)
    else:
        raise ValueError(f'不支持的 ASR 提供商: {Config.ASR_PROVIDER}')


def transcribe_whisper(audio_data: bytes) -> str:
    """使用 OpenAI Whisper API"""
    client = OpenAI(
        api_key=Config.OPENAI_API_KEY,
        base_url=Config.OPENAI_BASE_URL,
    )

    # 构造文件对象
    audio_file = ('audio.wav', audio_data, 'audio/wav')

    print('正在识别...')
    response = client.audio.transcriptions.create(
        model='whisper-1',
        file=audio_file,
        language='zh',
    )

    text = response.text.strip()
    print(f'识别结果: {text}')
    return text


if __name__ == '__main__':
    from recorder import record_until_silence
    audio = record_until_silence()
    text = transcribe(audio)
    print(f'\n你说的是: {text}')