SraVaani-1.0

Home/Models/ARTPARK-IISc/SraVaani-1.0

ARTPARK-IISc/SraVaani-1.0

kind: original
quant: bf16 (16-bit)
family: misc
modality: automatic-speech-recognition
root model: —
provider: ARTPARK-IISc
license: mit
language: hi, kn, ml, te, en, gu, pa, or, bn, ta, as, sa
context: —
library: —
vRAM (GB): —
size on disk: —
downloads: 1,322
likes: 18
created: —
last modified: 2026-07-31T06:45:20+00:00 (2026)
huggingface: https://huggingface.co/ARTPARK-IISc/SraVaani-1.0
tags: sravaani_tdtautomatic-speech-recognitioncustom_codehiknmlteengupaorbntaassanemrdataset:ARTPARK-IISc/Vaanidataset:ARTPARK-IISc/Vaani-transcription-partarxiv:2603.28714license:miteval-resultsregion:us

benchmark	category	score	conf	evidence
WER Common Voice	speech-asr	22.6	low	`\| Assamese \| 19.4 \| 22.6 \| 18.8 \| – \| – \| – \| – \| – \| 16.9 \|`
WER	speech-asr	19.4	low	`\| Assamese \| 19.4 \| 22.6 \| 18.8 \| – \| – \| – \| – \| – \| 16.9 \|`
WER FLEURS	speech-asr	18.8	low	`\| Assamese \| 19.4 \| 22.6 \| 18.8 \| – \| – \| – \| – \| – \| 16.9 \|`

none scraped.