SAVRN
Search Contact SAVRN

Open-weight model · Speech recognition

w2v-xls-r-uk

by Smoliakov Yehor/w2v-xls-r-uk

ATTENTION! Metrics (float16) using evaluate library with batchsize=1

Parameters315M
Context—
Weights1.3 GB
Licenseapache-2.0
AccessOpen weights
Monthly Downloads971.5k

Runs On

What it takes to serve w2v-xls-r-uk (315M parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them.

PrecisionWeightsMemory neededCheapest setupPer hourAlso fits
16-bit 0.6 GB 0.8 GB 1x MI300X (192 GB)
Vultr		 $1.85 1x H100 $1.99 · 1x MI325X $2.00
8-bit 0.3 GB 0.4 GB 1x MI300X (192 GB)
Vultr		 $1.85 1x H100 $1.99 · 1x MI325X $2.00
4-bit 0.2 GB 0.2 GB 1x MI300X (192 GB)
Vultr		 $1.85 1x H100 $1.99 · 1x MI325X $2.00

Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the SAVRN Index, read Sep 18, 2026.

SAVRN's Notes on w2v-xls-r-uk

Eight tenths of a gigabyte is all this speech recognizer needs at 16-bit, with the weights at 0.6 GB, and the 4-bit build drops to 0.2 GB. The cheapest Index host is one MI300X at $1.85 an hour, and 192 GB against a 0.8 GB model is a mismatch; run it beside other tenants. It is a 24-layer CTC design derived from facebook/wav2vec2-xls-r-300m.

Apache 2.0 allows commercial use, modification and redistribution, asks that license notices and any NOTICE file stay with the model and that significant changes be stated, and includes a patent grant. Access is open. Training data is mozilla-foundation/common_voice_10_0, and the publisher's own numbers on the uk split are a 3.64 CER and a 20.24 WER at float16, batch size 1. Run your own audio through it before committing; that is a single publisher-reported figure on a model released June 8, 2022.

SAVRN Research, 2026-09-18

Model Card

By Smoliakov, published under apache-2.0, revision e3ced4def0d7.

ATTENTION! Metrics (float16) using evaluate library with batchsize=1

Read Smoliakov's full model card

ATTENTION!

Use an updated model: https://huggingface.co/Yehor/w2v-bert-uk-v2.1

Community

  • Discord: https://bit.ly/discord-uds
  • Speech Recognition: https://t.me/speech_recognition_uk
  • Speech Synthesis: https://t.me/speech_synthesis_uk

See other Ukrainian models: https://github.com/egorsmkv/speech-recognition-uk

Evaluation results

Metrics (float16) using evaluate library with batch_size=1:

  • WER: 0.2024 metric, 20.24%
  • CER: 0.0364 metric, 3.64%
  • Accuracy on words: 79.76%
  • Accuracy on chars: 96.36%
  • Inference time: 63.4848 seconds
  • Audio duration: 16665.5212 seconds
  • RTF: 0.0038

Cite this work

@misc {smoliakov_2025,
    author       = { {Smoliakov} },
    title        = { w2v-xls-r-uk (Revision 55b6dc0) },
    year         = 2025,
    url          = { https://huggingface.co/Yehor/w2v-xls-r-uk },
    doi          = { 10.57967/hf/4556 },
    publisher    = { Hugging Face }
}

Configuration

Architecture
Wav2Vec2ForCTC
Layers
24
Hidden size
1,024
Feed-forward size
4,096
Attention heads
16
Vocabulary size
40
Stored precision
float32
Model type
wav2vec2

Identity and Version

Repository
Yehor/w2v-xls-r-uk
Publisher
Smoliakov
Task
Speech recognition
Modality
Audio
Library
transformers
Parameters
315M parameters
Languages
uk
Revision
e3ced4def0d70be3aab0f2db598a59961fe9ab3b
First published
2022-06-08
Last updated
2025-05-12

Files and Weights

13 files, 2.1 GB in total. The weights are 1 file totalling 1.3 GB in safetensors.

Weights1 file · 1.3 GB
Configuration6 files · 3.2 KB
Tokenizer2 files · 789 B
Documentation1 file · 1.6 KB
Other2 files · 853.8 MB
Repository1 file · 1.3 KB
Every file
FileTypeSizeSHA-256
model.safetensorsWeights1.3 GB e8d3ea382585
added_tokens.jsonConfiguration30 B —
alphabet.jsonConfiguration403 B —
config.jsonConfiguration2.1 KB —
language_model/attrs.jsonConfiguration78 B —
preprocessor_config.jsonConfiguration262 B —
special_tokens_map.jsonConfiguration406 B —
README.mdDocumentation1.6 KB —
language_model/lm.binaryOther848.0 MB f7b085a6a959
language_model/unigrams.txtOther5.8 MB —
.gitattributesRepository1.3 KB —
tokenizer_config.jsonTokenizer337 B —
vocab.jsonTokenizer452 B —

License and Download

License
apache-2.0
Access
Open weights, no gate
Download size
1.3 GB
Download from Smoliakov

Released by Smoliakov through its official repository on Hugging Face. Read the license.

Built From

  • Derived from facebook/wav2vec2-xls-r-300m
  • Trained on (disclosed) mozilla-foundation/common_voice_10_0

Evaluations

Each result is shown as reported, with the conditions its reporter stated. None is a SAVRN measurement. A comparison lines two results up only when their configuration, unit and setup are all stated and identical.

BenchmarkConditionsResultReported byRevisionDate
common_voice_10_0 Configuration ukTask Automatic Speech RecognitionMetric CERComparison conditions not established 3.64 Yehor
Publisher reported		 Evaluated revision not stated —
common_voice_10_0 Configuration ukTask Automatic Speech RecognitionMetric WERComparison conditions not established 20.24 Yehor
Publisher reported		 Evaluated revision not stated —

Memory Requirements

PrecisionWeights in memory
As published1.3 GB
16-bit0.6 GB
8-bit0.3 GB
4-bit0.2 GB

Weights only, from the published parameter count; the key-value cache and runtime add to this.

Questions About w2v-xls-r-uk

How much GPU memory does w2v-xls-r-uk need?

About 0.8 GB at 16-bit and 0.2 GB at 4-bit: the weights (315M parameters) plus a working margin. A long context needs more.

What is the cheapest GPU to run w2v-xls-r-uk on?

At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.

Can I use w2v-xls-r-uk commercially?

Yes. w2v-xls-r-uk is released under Apache License 2.0. The Apache License 2.0 is a permissive open-source license. It permits commercial use, modification and redistribution. It requires keeping the license and copyright notices and any NOTICE file, stating significant changes, and it includes an express patent grant from contributors.

Similar Models

Model · Speech recognition

romanian-wav2vec2

Théo Gigant

You can test this model online with the Space for Romanian Speech Recognition The model ranked TOP-1 on Romanian Speech Recognition during HuggingFace's Robust Speech Challenge: This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the Common Voice 8.0 - Romanian subset dataset, with extra training data from Romanian Speech Synthesis dataset. Without the 5-gram Language Model optimization, it achieves the following results on the evaluation set (Common Voice 8.0, Romanian subset, test split): The architecture is based on facebook/wav2vec2-xls-r-300m with a speech recognition CTC head and an added 5-gram language model (using pyctcdecode and kenlm) trained on the Romanian…

Open weights apache-2.0 315M parameters transformers
View model

Model · Speech recognition

wav2vec2-xls-r-300m-ftspeech

Dan Saattrup Smart

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the FTSpeech dataset, being a dataset of 1,800 hours of transcribed speeches from the Danish parliament. The model achieves the following WER scores (lower is better): The use of this model needs to adhere to this license from the Danish Parliament.

Open weights other 315M parameters transformers
View model

Finetuned version of KBs VoxRex large model using Swedish radio broadcasts, NST and Common Voice data. Evalutation without a language model gives the following: WER for NST + Common Voice test set (2% of total sentences) is 2.5%. WER for Common Voice test set is 8.49% directly and 7.37% with a 4-gram language model. When using this model, make sure that your speech input is sampled at 16kHz. This model has been fine-tuned for 120000 updates on NST + CommonVoice and then for an additional 20000 updates on CommonVoice only. The additional fine-tuning on CommonVoice hurts performance on the NST+CommonVoice test set somewhat and, unsurprisingly, improves it on the CommonVoice test set. It seems…

Open weights cc0-1.0 315M parameters transformers
View model

Model · Speech recognition

wav2vec2-xls-r-300m-cs-250

Vojtěch Drábek

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the commonvoice 8.0 dataset as well as other datasets listed below. It achieves the following results on the evaluation set: The eval.py script results using a LM are: Fine-tuned facebook/wav2vec2-large-xlsr-53 on Czech using the Common Voice dataset. When using this model, make sure that your speech input is sampled at 16kHz. The model can be used directly (without a language model) as follows: The model can be evaluated using the attached eval.py script: The Common Voice 8.0 train and validation datasets were used for training, as well as the following datasets: - Šmídl, Luboš and Pražák, Aleš, 2013, OVM – Otázky…

Open weights apache-2.0 315M parameters transformers
View model

Model · Speech recognition

wav2vec2-xls-r-300m-hebrew

Vladimir Gurevich

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the private datasets in 2 stages - firstly was fine-tuned on a small dataset with good samples Then the obtained model was fine-tuned on a large dataset with the small good dataset, with various samples from different sources, and with an unlabeled dataset that was weakly labeled using a previously trained model. (weakly labeled data wasn't used in validation set) on small dataset on large dataset on small dataset on large dataset The following hyperparameters were used during training: - learningrate: 0.0003 - trainbatchsize: 8 - evalbatchsize: 8 - distributedtype: multi-GPU - numdevices: 2 - gradientaccumulationsteps: 4…

Open weights 315M parameters transformers
View model

Model · Speech recognition

mms-300m-1130-forced-aligner

Mahmoud Ashraf

This Python package provides an efficient way to perform forced alignment between text and audio using Hugging Face's pretrained models. it also features an improved implementation to use much less memory than TorchAudio forced alignment API. The model checkpoint uploaded here is a conversion from torchaudio to HF Transformers for the MMS-300M checkpoint trained on forced alignment dataset

Open weights cc-by-nc-4.0 315M parameters transformers
View model