CoolFace
Datasetpublic

artmelancholy/golos_mfa_punctuation_long

Golos MFA Punctuation (Long) Long-form Russian speech derived from govnejri/golos_mfa_punctuation. Purpose Most public Russian STT corpora ship as short clips (a few seconds each). For benchmarking long-form transcription, VAD, punctuation, and streaming behavior, you want minutes-long audio with reliable word-level alignments. This dataset builds those long clips by splicing groups of consecutive short clips together, inserting randomized silences between them… See the full description on the dataset page: https://huggingface.co/datasets/artmelancholy/golos_mfa_punctuation_long.

sourceHugging Faceotherupdated 4mo agoView on Hugging Face
1likes18downloads
../
filegolos_mfa_punctuation_long_00000.parquet2.55 GBdownload

artmelancholy/golos_mfa_punctuation_long · main · files are served by the source, never re-hosted here