SAVRN
Search Contact SAVRN

Open-weight model · Speech recognition

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

by Vojtěch Drábek comodoro/wav2vec2-xls-r-300m-cs-250

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.

Parameters315M
Context—
Weights3.6 GB
Licenseapache-2.0
AccessOpen weights
Monthly Downloads1.3M

Runs On

What it takes to serve wav2vec2-xls-r-300m-cs-250 (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 wav2vec2-xls-r-300m-cs-250

Czech audio in, Czech text out, and the whole model fits in under a gigabyte. At 16-bit it needs 0.6 GB of weights and 0.8 GB of memory, so the cheapest listed setup, one MI300X with 192 GB at $1.85 an hour on-demand, would sit almost entirely idle under a single copy. We would pack many transcription streams onto that card instead. Budget disk separately: the files total 3.57 GB because the stored precision is float32.

Apache 2.0 allows commercial use, modification and redistribution provided the license and notices travel with it. Check the lineage before committing: it descends from facebook/wav2vec2-xls-r-300m and was trained on Common Voice 8.0, ovm, pscr and vystadial2016, and the publisher reports a 7.3 word error rate on Common Voice 8 against 38.5 on the other test data it lists. Feed it 16 kHz audio and test it on your own recordings.

SAVRN Research, 2026-09-18

Model Card

By Vojtěch Drábek, published under apache-2.0, revision 73e2b9004f35.

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

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the common_voice 8.0 dataset as well as other datasets listed below.

It achieves the following results on the evaluation set: - Loss: 0.1271 - Wer: 0.1475 - Cer: 0.0329

The eval.py script results using a LM are: - WER: 0.07274312090176113 - CER: 0.021207369275558875

Model description

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:

Read the full model card (550 words)

Configuration

Architecture
Wav2Vec2ForCTC
Layers
24
Hidden size
1,024
Feed-forward size
4,096
Attention heads
16
Vocabulary size
46
Stored precision
float32
Model type
wav2vec2

Identity and Version

Repository
comodoro/wav2vec2-xls-r-300m-cs-250
Publisher
Vojtěch Drábek
Task
Speech recognition
Modality
Audio
Library
transformers
Parameters
315M parameters
Languages
cs
Revision
73e2b9004f35d3ca79316a624e7edc3cdaa45d40
First published
2022-03-02
Last updated
2023-10-31

Files and Weights

19 files, 3.6 GB in total. The weights are 4 files totalling 3.6 GB in bin, pth, safetensors.

Weights4 files · 3.6 GB
Configuration8 files · 11.4 KB
Tokenizer2 files · 769 B
Documentation1 file · 7.2 KB
Other3 files · 5.5 MB
Repository1 file · 1.2 KB
Every file
FileTypeSizeSHA-256
language_model/cs-500k.binWeights1.0 GB 8d8d4a962c06
model.safetensorsWeights1.3 GB 99376277738c
pytorch_model.binWeights1.3 GB ef2fdad28a59
rng_state.pthWeights14.6 KB 490ca80b725a
added_tokens.jsonConfiguration23 B —
alphabet.jsonConfiguration343 B —
config.jsonConfiguration2.1 KB —
eval.pyConfiguration5.6 KB —
language_model/attrs.jsonConfiguration78 B —
preprocessor_config.jsonConfiguration262 B —
special_tokens_map.jsonConfiguration309 B —
trainer_state.jsonConfiguration2.8 KB —
README.mdDocumentation7.2 KB —
Fine_Tune_XLS_R_on_Common_Voice_cs_300m_CV8.ipynbOther361.7 KB —
language_model/unigrams.txtOther5.0 MB —
train.ipynbOther161.2 KB —
.gitattributesRepository1.2 KB —
tokenizer_config.jsonTokenizer300 B —
vocab.jsonTokenizer469 B —

License and Download

License
apache-2.0
Access
Open weights, no gate
Download size
3.6 GB
Download from Vojtěch Drábek

Released by Vojtěch Drábek 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_8_0
  • Trained on (disclosed) ovm
  • Trained on (disclosed) pscr
  • Trained on (disclosed) vystadial2016

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 8 Task Automatic Speech RecognitionMetric Test CERComparison conditions not established 2.1 comodoro
Publisher reported		 Evaluated revision not stated —
Common Voice 8 Task Automatic Speech RecognitionMetric Test WERComparison conditions not established 7.3 comodoro
Publisher reported		 Evaluated revision not stated —
Robust Speech Event - Dev Data Task Automatic Speech RecognitionMetric Test WERComparison conditions not established 43.44 comodoro
Publisher reported		 Evaluated revision not stated —
Robust Speech Event - Test Data Task Automatic Speech RecognitionMetric Test WERComparison conditions not established 38.5 comodoro
Publisher reported		 Evaluated revision not stated —

Memory Requirements

PrecisionWeights in memory
As published3.6 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 wav2vec2-xls-r-300m-cs-250

How much GPU memory does wav2vec2-xls-r-300m-cs-250 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 wav2vec2-xls-r-300m-cs-250 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 wav2vec2-xls-r-300m-cs-250 commercially?

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

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

This model for Croatian ASR is based on the facebook/wav2vec2-xls-r-300m model and was fine-tuned with 300 hours of recordings and transcripts from the ASR Croatian parliament dataset ParlaSpeech-HR v1.0. If you use this model, please cite the following paper: Nikola Ljubešić, Danijel Koržinek, Peter Rupnik, Ivo-Pavao Jazbec. ParlaSpeech-HR -- a freely available ASR dataset for Croatian bootstrapped from the ParlaMint corpus. http://www.lrec-conf.org/proceedings/lrec2022/workshops/ParlaCLARINIII/pdf/2022.parlaclariniii-1.16.pdf Evaluation is performed on the dev and test portions of the ParlaSpeech-HR v1.0 dataset. There are multiple models available, and in terms of CER and WER, the…

Open weights 315M parameters transformers
View model

Model · Speech recognition

w2v-xls-r-uk

Smoliakov

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

Open weights apache-2.0 315M parameters transformers
View model

Model · Speech recognition

koelsch-wav2vec2-ipa

Chem

IPA phoneme recognition for Kölsch (Ripuarian German, Cologne). Fine-tuned from facebook/wav2vec2-xls-r-300m with a CTC head over a 48-symbol IPA inventory. This is also the forced-alignment model. Its frame-level CTC posteriors drive torchaudio.functional.forcedalign to produce Praat TextGrids with word and phone tiers — see notebook 9 in the repository. Trained during the CIF Tandem Fellowship at IfL-Phonetik, University of Cologne, on Alles Kölsch (Bhatt & Lindlar 1998) — 4,670 utterances, 4.5 hours of spontaneous Cologne dialect speech from 105 speakers aged 10–88 across 49 neighbourhoods. Kölsch is Ripuarian German. It has no public speech dataset and no standardised spelling: 94.1 %…

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

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-large-xls-r-300m-Urdu

Abid Ali Awan

A fine-tuned XLS-R 300M CTC model for Urdu automatic speech recognition. It transcribes 16 kHz mono audio and includes an optional 5-gram KenLM decoder. Best reported result: 39.89% WER / 16.70% CER with KenLM decoding on the Urdu Common Voice 8.0 test set. See the Kaggle evaluation notebook for a reproducible example. The repository contains a 5-gram KenLM language model. The Kaggle notebook evaluates a five-sample streaming smoke test from fixie-ai/commonvoice170 (ur, test). Results are reported on the Urdu test split of Mozilla Common Voice 8.0. The language-model row is the model-card score; compare each result only with the same decoding strategy. To reproduce language-model evaluation…

Open weights apache-2.0 315M parameters transformers
View model