hviske-v5

Home/Models/syvai/hviske-v5

syvai/hviske-v5

kind: original
quant: bf16 (16-bit)
family: misc
modality: automatic-speech-recognition
root model: —
provider: syvai
license: cc-by-nc-4.0
language: da
context: —
library: —
vRAM (GB): —
size on disk: —
downloads: 189
likes: 1
created: —
last modified: 2026-05-21T13:33:12+00:00 (2026)
huggingface: https://huggingface.co/syvai/hviske-v5
tags: safetensorscohere_asrautomatic-speech-recognitiondanishwhispercustom_codedadataset:syvai/danish-asr-unifieddataset:nvidia/Granarydataset:alexandrainst/notadataset:alexandrainst/ftspeechdataset:CoRal-project/coral-v3dataset:alexandrainst/nst-dadataset:fixie-ai/common_voice_17_0license:cc-by-nc-4.0region:us

benchmark	category	score	context	conf	evidence
WER Common Voice	speech-asr	24.3	Common Voice 17	medium	`\| Common Voice 17 \| Crowd-sourced \| 24.3% \| 7.4% \|`
WER Common Voice	speech-asr	17.0		medium	`Common Voice 17`
WER	speech-asr	11.8		low	`\| VoxPopuli \| European Parliament \| 11.8% \| 6.3% \|`
WER VoxPopuli	speech-asr	11.8	VoxPopuli	medium	`\| VoxPopuli \| European Parliament \| 11.8% \| 6.3% \|`
CER	speech-asr	6.3		low	`\| VoxPopuli \| European Parliament \| 11.8% \| 6.3% \|`

none scraped.