Dataset · parler-tts
Filtered 585-hour English speech corpus
parler-tts/libritts_r_filtered
A filtered version of the LibriTTS-R audio dataset with improved sound quality for speech synthesis and recognition.
What's inside
The dataset contains pairs of English reading audio recordings (24 kHz) and text transcripts. The original LibriTTS-R corpus contains about 585 hours of speech from multiple speakers, and the current sample size ranges from 100,000 to 1 million records. Samples with poor audio restoration and speakers with overlapping voices were removed from the data.
What you can do with it
- Training and fine-tuning text-to-speech (TTS) models with clean audio
- Training English speech recognition (ASR) systems
- Voice cloning and creating multi-speaker voice models
- Tasks
- Text → speech, Automatic speech recognition
- Languages
- English
- Size
- 100K–1M rows
- Downloads this month
- 4.3K
- Likes
- 26
Models trained on it
- remsky/kokoro-inno-clone-tunerText → speech 1.7M
- parler-tts/parler-tts-mini-v1Text → speech 31.5K
- parler-tts/parler-tts-mini-multilingual-v1.1Text → speech 12.6K
- parler-tts/parler-tts-mini-v1.1Text → speech 5.8K
- parler-tts/parler-tts-large-v1Text → speech 4.8K
- OuteAI/OuteTTS-0.2-500M-GGUFText → speech 2.9K
- OuteAI/OuteTTS-0.1-350M-GGUFText → speech 748
- cstr/Sidon-GGUFAudio → audio 479
How to load it
from datasets import load_dataset
ds = load_dataset("parler-tts/libritts_r_filtered")