Voila-Tokenizer

Home/Models/maitrix-org/Voila-Tokenizer

maitrix-org/Voila-Tokenizer

kind: original
quant: bf16 (16-bit)
family: misc
modality: audio-to-audio
root model: —
provider: maitrix-org
license: mit
language: en, zh, fr, de, ja, ko
context: —
library: transformers
vRAM (GB): —
size on disk: —
downloads: 294
likes: 9
created: —
last modified: 2025-05-06T14:51:20+00:00 (2025)
huggingface: https://huggingface.co/maitrix-org/Voila-Tokenizer
tags: transformerssafetensorsencodecfeature-extractionaudio-to-audioenzhfrdejakodataset:maitrix-org/Voila-Benchmarkdataset:maitrix-org/Voila-million-voicearxiv:2505.02707license:mitendpoints_compatibleregion:us

benchmark	category	score	context	conf	evidence
WER LibriSpeech	speech-asr	4.8	Voila (w/o LibriSpeech train split)	medium	`\| Voila (w/o LibriSpeech train split) \| 4.8 \|`
WER	speech-asr	2.7		low	`\| Whisper large v2 \| 2.7 \|`
WER LibriSpeech	speech-asr	2.7	Voila (with LibriSpeech train split)	medium	`\| Voila (with LibriSpeech train split) \| 2.7 \|`

none scraped.