| from .vad import AudioVad, AudioSegment | |
| import numpy as np | |
| from typing import List, Tuple | |
| class VoiceActivityDetector: | |
| """ | |
| VAD检测器，用于检测音频中的语音片段 | |
| 这是一个包装类，内部使用AudioVad实现功能 | |
| """ | |
| def __init__(self, save_audio=True, save_json=True): | |
| """ | |
| 初始化VAD检测器 | |
| :param save_audio: 是否保存分段音频 | |
| :param save_json: 是否保存JSON元数据 | |
| """ | |
| self.vad = AudioVad( | |
| save_audio=save_audio, | |
| save_json=save_json, | |
| output_dir="dataset/audio/segments", | |
| json_dir="dataset/audio/metadata" | |
| ) | |
| def detect_voice_segments(self, audio_data: np.ndarray, sample_rate: int) -> List[Tuple[float, float]]: | |
| """ | |
| 检测音频中的语音片段 | |
| :param audio_data: 音频数据 | |
| :param sample_rate: 采样率 | |
| :return: 语音片段列表，每个片段为(开始时间, 结束时间)的元组 | |
| """ | |
| segments = self.vad.process_audio_data(audio_data, sample_rate) | |
| return [(segment.start_time, segment.end_time) for segment in segments] | |
| __all__ = ['AudioVad', 'AudioSegment', 'VoiceActivityDetector'] | |