robmsmt/ASR-Audio-Data-Links

Organizes diverse publicly and commercially available speech datasets across multiple languages and acoustic conditions in a single curated table, including conversational, read, lecture, and noisy speech corpora ranging from 5 hours (TIMIT) to 33,000+ hours (GigaSpeech). Covers both ASR training data and text-to-speech (TTS) datasets with direct download links, torrent sources, and LDC catalog codes, enabling researchers to compare dataset sizes and characteristics for benchmark evaluation.

231 stars. No commits in the last 6 months.