Sova dataset
Key facts: - Russian, English and Chinese languages - ~ 32 328 hours - ~ 3,21 TB in `.wav` format The project is distributed under the Other license, first published in 2019. Key topics include: audio, audio-data, audio-dataset, audio-datasets, chinese-dataset.
SOVA Dataset
SOVA Dataset is free public STT/ASR dataset.
Key facts:
- Russian, English and Chinese languages
- ~ 32 328 hours
- ~ 3,21 TB in
.wavformat
Dataset composition
| Name | Lang | Hours | Size | Source | Equipment | Annotation | Speech type | Augmentation | Quality | |
|---|---|---|---|---|---|---|---|---|---|---|
| EngAudiobooksOriginal | Download | EN | 7 130 | 743 Gb | audiobook | professional | forced alignment | reading | none | 95% |
| EngAudiobooksNoisy | Download | EN | 3 873 | 310 Gb | audiobook | professional | forced alignment | reading | phone calls | 95% |
| RuAudiobooksDevices | Download | RU | 298 | 30,24 Gb | audiobook | unprofessional | manual | reading | none | 99% |
| RuDevices | Download | RU | 101 | 10,42 Gb | audio records | unprofessional | manual | live speech | none | 98% |
| RuYoutube | Download | RU | 17 451 | 1 873 Gb | audio records | unprofessional | asr | live speech | none | 95% |
| ZhYoutube | Download | CN | 3 475,1 | 321 Gb | audio records | unprofessional | asr | live speech | none | 97.83% |
| TOTAL | - | - | 32 328,1 | 3 287,66 Gb<br>(3,21 TB) | - | - | - | - | - | - |
Audio characteristics
- Bit rate mode: constant
- Bit rate: 256 kbps
- Channel(s): 1 channel
- Sample rate: 16.0 kHz
- Bit depth: 16 bit
Updates
- 08/11/2022: Release v0.4.0
- 10/12/2021: Release v0.3.0
- 22/12/2020: Release v0.2.0
- 24/12/2019: Published dataset with 116 hours.
Contacts
For all questions please feel free to contact us <a href="mailto:support@sova.ai?subject=SOVA Dataset">support@sova.ai</a>
License
SOVA Dataset is licensed under Creative Commons BY 4.0 license by Virtual Assistant, LLC.
Contributors
Showing top 2 contributors by commit count.
Related Repositories
huggingface/transformers
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.
FFmpeg/FFmpeg
Mirror of https://git.ffmpeg.org/ffmpeg.git
mpv-player/mpv
🎥 Command line media player
OpenBMB/VoxCPM
VoxCPM2: Tokenizer-Free TTS for Multilingual Speech Generation, Creative Voice Design, and True-to-Life Cloning
ossrs/srs
SRS is a simple, high-efficiency, real-time media server supporting RTMP, WebRTC, HLS, HTTP-FLV, HTTP-TS, SRT, MPEG-DASH, and GB28181, with codec support for H.264, H.265, AV1, VP9, AAC, Opus, and G.711.
goldfire/howler.js
Javascript audio library for the modern web.