LibriVAD Small: Noise-Augmented Speech Dataset

A 15GB variant of the LibriVAD dataset, which is built on the LibriSpeech corpus. The dataset is noise-augmented, suggesting it is designed for training models in noisy acoustic environments. Its author, organization, and specific creation date are unknown.

Use Cases

• Train voice activity detection models based on the noise-augmented speech data.
• Benchmark speech recognition systems in noisy conditions based on the described augmentation.
• Develop audio denoising or enhancement algorithms using the augmented LibriSpeech samples.

Strengths

• 15GB size provides a substantial volume of audio data for model training.
• Built on the established LibriSpeech corpus, which is a known benchmark in speech research.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Last update date is unknown; freshness unverified.

Provenance

Source

Kaggle

Collection Method

Derived and augmented from the LibriSpeech corpus.