LibriTTS-R Filtered: A Multi-Speaker English Speech Corpus

Filtered LibriTTS-R is a multi-speaker English speech corpus of approximately 585 hours of read speech at 24kHz sampling rate. It is a filtered version of the LibriTTS-R corpus, excluding samples flagged for failed speech restoration and speakers with detected multi-speaker issues. The dataset was created by parler-tts and last updated on August 6, 2024.

Use Cases

• Training speech synthesis models based on high-quality, 24kHz audio samples.
• Evaluating speech restoration algorithms based on a corpus filtered for problematic samples.
• Developing multi-speaker voice systems based on a corpus of read English speech.
• Benchmarking audio quality improvements based on a filtered version of a known corpus.

Strengths

• Approximately 585 hours of audio data provides substantial volume for training.
• Filtering based on two published sources likely improves data quality by removing problematic samples.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and file formats are unknown, which may limit suitability assessment.
• The description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

Filtered version of the LibriTTS-R corpus.

Collection Method

Filtering based on lists from the LibriTTS-R paper and LibriTTS-P.

Freshness

Last updated 2024-08-06 16:45:54; freshness should be verified.