voa-engines/vad-dataset-comparison

VOA PT-BR VAD比较数据集包含使用三种不同语音活动检测(VAD)预设处理的音频段：in_streaming_vad、Old_streaming_vad和aggressive_vad。每个音频段都有.wav格式的音频文件和包括 Whisper自动转录、持续时间、开始和结束时间以及VAD参数的元数据。该数据集用于评估VAD参数、比较不同VAD配置，以及葡萄牙语自动语音识别(ASR)模型的训练和微调。

The VOA PT-BR VAD Comparison Dataset contains audio segments processed with three different Voice Activity Detection (VAD) presets: in_streaming_vad, Old_streaming_vad, and aggressive_vad. Each segment includes a .wav audio file and associated metadata with automated transcriptions by Whisper, duration, start and end times, and the VAD parameters used. This dataset is intended for evaluating VAD parameters, comparing different VAD configurations, and training and fine-tuning Portuguese Automatic Speech Recognition (ASR) models.