Dataset · mythicinfinity
Enhanced English speech corpus for TTS (585 hours)
mythicinfinity/libritts_r
About 585 hours of clean multi-speaker English speech (24 kHz) with texts for speech synthesis.
What's inside
LibriTTS-R is an updated version of the 2019 LibriTTS corpus with improved audio quality. The dataset contains between 100K and 1M pairs of English audio and text with a total duration of about 585 hours, recorded by different speakers at a 24 kHz sampling rate and split into 7 parts.
What you can do with it
- Train and fine-tune multi-speaker text-to-speech (TTS) models
- Train vocoders and voice cloning systems on clean 24 kHz audio
- Tasks
- Text → speech
- Languages
- English
- Size
- 100K–1M rows
- Downloads this month
- 17.8K
- Likes
- 49
Models trained on it
- neuphonic/neucodecAudio → audio 55K
- SPRINGLab/Indic-MioText → speech 4K
- eugenehp/neucodecAudio → audio 852
- neuphonic/distill-neucodecAudio → audio 692
- nguyensu27/neucodecAudio → audio 301
- mmmhusto/TTS-EnglishText → speech 166
- shangeth/Wren-TTS-0.5B-multi-expressiveText → speech 155
- ibm-research/DAC.speech.v1.0 129
How to load it
from datasets import load_dataset
ds = load_dataset("mythicinfinity/libritts_r")