higgs-audio-v3-8b-stt-v2

Home/Models/bosonai/higgs-audio-v3-8b-stt-v2

bosonai/higgs-audio-v3-8b-stt-v2

kind: original
quant: bf16 (16-bit)
family: misc
modality: automatic-speech-recognition
root model: bosonai/higgs-audio-v3-8b
provider: bosonai
license: apache-2.0
language: en
context: —
library: —
vRAM (GB): —
size on disk: —
downloads: 1,134
likes: 14
created: —
last modified: 2026-05-22T18:16:30+00:00 (2026)
huggingface: https://huggingface.co/bosonai/higgs-audio-v3-8b-stt-v2
tags: safetensorshiggs_audio_3automatic-speech-recognitionhf-asr-leaderboardwhisperqwencustom_codeenlicense:apache-2.0model-indexregion:us

benchmark	category	score	context	conf	evidence
WER	speech-asr	10.14		low	`\| AMI \| 10.14% \|`
WER AMI	speech-asr	10.14	AMI	medium	`\| AMI \| 10.14% \|`
WER Earnings-22	speech-asr	8.73	Earnings22	medium	`\| Earnings22 \| 8.73% \|`
WER GigaSpeech	speech-asr	8.47	GigaSpeech	medium	`\| GigaSpeech \| 8.47% \|`
WER VoxPopuli	speech-asr	5.92	VoxPopuli	medium	`\| VoxPopuli \| 5.92% \|`
WER SPGISpeech	speech-asr	3.6	SPGISpeech	medium	`\| SPGISpeech \| 3.60% \|`
WER TED-LIUM	speech-asr	3.09	TED-LIUM	medium	`\| TED-LIUM \| 3.09% \|`
WER LibriSpeech	speech-asr	2.38	LibriSpeech Other	medium	`\| LibriSpeech Other \| 2.38% \|`
WER LibriSpeech	speech-asr	1.25	LibriSpeech Clean	medium	`\| LibriSpeech Clean \| 1.25% \|`

none scraped.