3 datasets found

MLS: A large-scale multilingual dataset for speech research.

The sequence-to-sequence approach is widely used in speech recognition (SR) nowadays, and many research works are dedicated to show that their capabilities relying on a single...

The dictation dataset across 39 locales, including Latin (Albanian, Icelandic, Slovak), Arabic (Levant, Maghrebi), Cyrillic (Macedonian, Kazakh), Devanagari (Nepali), etc.