相关数据集
ASR-CCantCSC: A Chinese Cantonese (Canton) Conversational Speech Corpus
This open-source dataset consists of 4.25 hours of transcribed Guangzhou Cantonese conversational speech on certain topics, where ten conversations between ten pairs of speakers were contained.
MagicHub开源社区2021-03-18 更新340
kjetMol/ArtificiallyNoisySpeechTranscriptions
该数据集包含来自Språkbanken提供的挪威语语料库中nb_samtale子集的语音文件转录。这些转录文件经过了受控的噪声添加，以模拟不同的声学环境。原始音频的持续时间为24秒到27秒，格式为WAV，共有9个文件。转录部分测试了3个模型、4种噪声类型和16个噪声级别，总文件数为1682个。单词错误率（WER）用于评估语音识别系统在不同噪声条件下的性能，计算基于0%噪声添加的转录作为基准。
Hugging Face2024-05-03 更新200
Wake Word Indian English Dataset
This dataset features a rich array of audio recordings for wake word detection in Indian English. It encompasses diverse accents, regional dialects, and recording conditions to enhance the performance
ms.shaip.com2024-07-26 更新230
atgarcia/trainDataset5
--- dataset_info: features: - name: text dtype: string - name: audio struct: - name: array sequence: float64 - name: path dtype: string - name: sampling_rate
Hugging Face2024-02-26 更新100
