- WSJ0(Wall Street Journal)
- 可用于语音分离、语音增强、语音识别等任务 for tasks including Speech Separation, Speech Enhancement, ASR...
- paper (HLT 1992)
- LDC (官方链接 official link)
- LibriSpeech
- 可用于语音识别等任务 for tasks including ASR, Speech Separation, Speech Enhancement...
- paper (ICASSP 2015)
- OpenSLR (官方链接 official link)
- RIR_NOISES
- 房间冲击响应(RIR)数据, 可用于语音识别、说话人识别中的数据增强 Room Impulse Response(RIR) data for data augmentation in ASR, Speaker Recognition...
- paper (ICASSP 2017)
- OpenSLR (官方链接 official link)
- wsj0-2mix
- 可用于语音分离、语音增强等任务 for tasks including Speech Separation, Speech Enhancement...
- paper (ICASSP 2016)
- **大学李宏毅博士上传的部分数据 some data released by Dr. Hung-yi Lee @ NTU
- website@MERL (MERL的数据生成官方脚本 official data generation scripts from MERL)
- LibriMix
- 可用于语音分离、语音增强等任务 for tasks including Speech Separation, Speech Enhancement...
- paper (arXiv 2020)
- GitHub (Inria的数据生成官方脚本 official data generation scripts from Inria
- WHAM!
- 噪声数据, 可用于语音增强、有噪语音分离等任务(配合wsj0-2mix、LibriMix中的语音数据) noise data for tasks including Speech Enhancement, Speech Separation under noisy environment...(normally used with speech audio from wsj0-2mix or LibriMix)
- paper (INTERSPEECH 2019)
- [email protected] (官方链接 official link)
- VoxSRC-20 track 1
- 说话人识别竞赛, 只能用VoxCeleb数据, 在VoxCeleb上进行评测 Speaker Recognition competition evaluated on VoxCeleb, only VoxCeleb can be used
- website@VoxSRC-20 (竞赛官网 challenge website)
- Leaderboard
- VoxSRC-20 track 2
- 说话人识别竞赛, 无数据约束, 在VoxCeleb上进行评测 Speaker Recognition competition evaluated on VoxCeleb, no limitation on data
- website@VoxSRC-20 (竞赛官网 challenge website)
- Leaderboard
- VoxCeleb1
- 可用于说话人验证、说话人辨认等任务 for tasks including Speaker Verification, Speaker Identification...
- paper (INTERSPEECH 2017)
- website@VGG-Oxford (官方链接 official link)
- website@Graviti
- VoxCeleb2
- 可用于说话人验证、说话人辨认等任务 for tasks including Speaker Verification, Speaker Identification...
- paper (INTERSPEECH 2018)
- website@VGG-Oxford (官方链接 official link)
- website@Graviti
- CN-Celeb1
- 可用于说话人验证、说话人辨认、说话人检索等任务 for tasks including Speaker Verification, Speaker Identification, Speaker Retrieval...
- paper (ICASSP 2020)
- OpenSLR (官方链接 official link)
- CN-Celeb2
- 可用于说话人验证、说话人辨认、说话人检索等任务 for tasks including Speaker Verification, Speaker Identification, Speaker Retrieval...
- paper (Speech Communication 2022)
- OpenSLR (官方链接 official link)
- wq2012/awesome-diarization
- VoxSRC-20 track 4
- 说话人分割竞赛, 无数据约束, 在VoxConverse上进行评测 Speaker Diariaztion competition evaluated on VoxConverse, no limitation on data
- website@VoxSRC-20 (竞赛官网 challenge website)
- Leaderboard
- VoxConverse
- 可用于说话人分割相关任务 for speaker diarization and related tasks...
- paper (INTERSPEECH 2020)
- website@VGG-Oxford (官方链接 official link)
- GitHub (VGG-Oxford的标签数据 ground-truth labels from VGG-Oxford)
- Spectral Clustering for Speaker Diariazation
- VoxSRC-20 Evaluation Toolkits
- 包括DER、JER等指标 metrics including DER, JER...
- GitHub (VoxSRC-20的官方程序 official code from VoxSRC-20)