SAVRN
Search Contact SAVRN

Open-weight model · Speech recognition

wav2vec2-large-xlsr-53-russian

by Jonatas Grosman jonatasgrosman/wav2vec2-large-xlsr-53-russian

Fine-tuned facebook/wav2vec2-large-xlsr-53 on Russian using the train and validation splits of Common Voice 6.1 and CSS10. When using this model, make sure that your speech input is sampled at 16kHz.

Parameters
Context
Weights2.5 GB
Licenseapache-2.0
AccessOpen weights
Monthly Downloads4.2M

Model Card

By Jonatas Grosman, published under apache-2.0, revision 232910050889.

Fine-tuned facebook/wav2vec2-large-xlsr-53 on Russian using the train and validation splits of Common Voice 6.1 and CSS10. When using this model, make sure that your speech input is sampled at 16kHz. This model has been fine-tuned thanks to the GPU credits generously given by the OVHcloud:) The script used for training can be found here: https://github.com/jonatasgrosman/wav2vec2-sprint The model can be used directly (without a language model) as follows... Using the HuggingSound library: 1. To evaluate on mozilla-foundation/commonvoice60 with split test 2. To evaluate on speech-recognition-community-v2/devdata If you want to cite this model you can use this

Read Jonatas Grosman's full model card

Fine-tuned XLSR-53 large model for speech recognition in Russian

Fine-tuned facebook/wav2vec2-large-xlsr-53 on Russian using the train and validation splits of Common Voice 6.1 and CSS10. When using this model, make sure that your speech input is sampled at 16kHz.

This model has been fine-tuned thanks to the GPU credits generously given by the OVHcloud :)

The script used for training can be found here: https://github.com/jonatasgrosman/wav2vec2-sprint

Usage

The model can be used directly (without a language model) as follows...

Using the HuggingSound library:

from huggingsound import SpeechRecognitionModel

model = SpeechRecognitionModel("jonatasgrosman/wav2vec2-large-xlsr-53-russian")
audio_paths = ["/path/to/file.mp3", "/path/to/another_file.wav"]

transcriptions = model.transcribe(audio_paths)

Writing your own inference script:

import torch
import librosa
from datasets import load_dataset
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor

LANG_ID = "ru"
MODEL_ID = "jonatasgrosman/wav2vec2-large-xlsr-53-russian"
SAMPLES = 5

test_dataset = load_dataset("common_voice", LANG_ID, split=f"test[:{SAMPLES}]")

processor = Wav2Vec2Processor.from_pretrained(MODEL_ID)
model = Wav2Vec2ForCTC.from_pretrained(MODEL_ID)

# Preprocessing the datasets.
# We need to read the audio files as arrays
def speech_file_to_array_fn(batch):
    speech_array, sampling_rate = librosa.load(batch["path"], sr=16_000)
    batch["speech"] = speech_array
    batch["sentence"] = batch["sentence"].upper()
    return batch

test_dataset = test_dataset.map(speech_file_to_array_fn)
inputs = processor(test_dataset["speech"], sampling_rate=16_000, return_tensors="pt", padding=True)

with torch.no_grad():
    logits = model(inputs.input_values, attention_mask=inputs.attention_mask).logits

predicted_ids = torch.argmax(logits, dim=-1)
predicted_sentences = processor.batch_decode(predicted_ids)

for i, predicted_sentence in enumerate(predicted_sentences):
    print("-" * 100)
    print("Reference:", test_dataset[i]["sentence"])
    print("Prediction:", predicted_sentence)
Reference Prediction
ОН РАБОТАТЬ, А ЕЕ НЕ УДЕРЖАТЬ НИКАК — БЕГАЕТ ЗА КЛЁШЕМ КАЖДОГО БУЛЬВАРНИКА. ОН РАБОТАТЬ А ЕЕ НЕ УДЕРЖАТ НИКАК БЕГАЕТ ЗА КЛЕШОМ КАЖДОГО БУЛЬБАРНИКА
ЕСЛИ НЕ БУДЕТ ВОЗРАЖЕНИЙ, Я БУДУ СЧИТАТЬ, ЧТО АССАМБЛЕЯ СОГЛАСНА С ЭТИМ ПРЕДЛОЖЕНИЕМ. ЕСЛИ НЕ БУДЕТ ВОЗРАЖЕНИЙ Я БУДУ СЧИТАТЬ ЧТО АССАМБЛЕЯ СОГЛАСНА С ЭТИМ ПРЕДЛОЖЕНИЕМ
ПАЛЕСТИНЦАМ НЕОБХОДИМО СНАЧАЛА УСТАНОВИТЬ МИР С ИЗРАИЛЕМ, А ЗАТЕМ ДОБИВАТЬСЯ ПРИЗНАНИЯ ГОСУДАРСТВЕННОСТИ. ПАЛЕСТИНЦАМ НЕОБХОДИМО СНАЧАЛА УСТАНОВИТЬ С НИ МИР ФЕЗРЕЛЕМ А ЗАТЕМ ДОБИВАТЬСЯ ПРИЗНАНИЯ ГОСУДАРСТВЕНСКИ
У МЕНЯ БЫЛО ТАКОЕ ЧУВСТВО, ЧТО ЧТО-ТО ТАКОЕ ОЧЕНЬ ВАЖНОЕ Я ПРИБАВЛЯЮ. У МЕНЯ БЫЛО ТАКОЕ ЧУВСТВО ЧТО ЧТО-ТО ТАКОЕ ОЧЕНЬ ВАЖНОЕ Я ПРЕДБАВЛЯЕТ
ТОЛЬКО ВРЯД ЛИ ПОЙМЕТ. ТОЛЬКО ВРЯД ЛИ ПОЙМЕТ
ВРОНСКИЙ, СЛУШАЯ ОДНИМ УХОМ, ПЕРЕВОДИЛ БИНОКЛЬ С БЕНУАРА НА БЕЛЬ-ЭТАЖ И ОГЛЯДЫВАЛ ЛОЖИ. ЗЛАЗКИ СЛУШАЮ ОТ ОДНИМ УХАМ ТЫ ВОТИ В ВИНОКОТ СПИЛА НА ПЕРЕТАЧ И ОКЛЯДЫВАЛ БОСУ
К СОЖАЛЕНИЮ, СИТУАЦИЯ ПРОДОЛЖАЕТ УХУДШАТЬСЯ. К СОЖАЛЕНИЮ СИТУАЦИИ ПРОДОЛЖАЕТ УХУЖАТЬСЯ
ВСЁ ЖАЛОВАНИЕ УХОДИЛО НА ДОМАШНИЕ РАСХОДЫ И НА УПЛАТУ МЕЛКИХ НЕПЕРЕВОДИВШИХСЯ ДОЛГОВ. ВСЕ ЖАЛОВАНИЕ УХОДИЛО НА ДОМАШНИЕ РАСХОДЫ И НА УПЛАТУ МЕЛКИХ НЕ ПЕРЕВОДИВШИХСЯ ДОЛГОВ
ТЕПЕРЬ ДЕЛО, КОНЕЧНО, ЗА ТЕМ, ЧТОБЫ ПРЕВРАТИТЬ СЛОВА В ДЕЛА. ТЕПЕРЬ ДЕЛАЮ КОНЕЧНО ЗАТЕМ ЧТОБЫ ПРЕВРАТИТЬ СЛОВА В ДЕЛА
ДЕВЯТЬ ЛЕВЕТЬ

Evaluation

  1. To evaluate on mozilla-foundation/common_voice_6_0 with split test
python eval.py --model_id jonatasgrosman/wav2vec2-large-xlsr-53-russian --dataset mozilla-foundation/common_voice_6_0 --config ru --split test
  1. To evaluate on speech-recognition-community-v2/dev_data
python eval.py --model_id jonatasgrosman/wav2vec2-large-xlsr-53-russian --dataset speech-recognition-community-v2/dev_data --config ru --split validation --chunk_length_s 5.0 --stride_length_s 1.0

Citation

If you want to cite this model you can use this:

@misc{grosman2021xlsr53-large-russian,
  title={Fine-tuned {XLSR}-53 large model for speech recognition in {R}ussian},
  author={Grosman, Jonatas},
  howpublished={\url{https://huggingface.co/jonatasgrosman/wav2vec2-large-xlsr-53-russian}},
  year={2021}
}

Configuration

Architecture
Wav2Vec2ForCTC
Layers
24
Hidden size
1,024
Feed-forward size
4,096
Attention heads
16
Vocabulary size
39
Model type
wav2vec2

Identity and Version

Repository
jonatasgrosman/wav2vec2-large-xlsr-53-russian
Publisher
Jonatas Grosman
Task
Speech recognition
Modality
Audio
Library
transformers
Parameters
Not stated by the source
Languages
ru
Revision
2329100508896c6d9b157019803ab5601e6f3406
First published
2022-03-02
Last updated
2022-12-14

Files and Weights

24 files, 4.0 GB in total. The weights are 2 files totalling 2.5 GB in bin, msgpack.

Weights2 files · 2.5 GB
Configuration6 files · 8.7 KB
Tokenizer1 file · 387 B
Documentation1 file · 6.9 KB
Other13 files · 1.5 GB
Repository1 file · 736 B
Every file
FileTypeSizeSHA-256
flax_model.msgpackWeights1.3 GB 967b19d64006
pytorch_model.binWeights1.3 GB d1cdb1a7921d
alphabet.jsonConfiguration263 B
config.jsonConfiguration1.8 KB
eval.pyConfiguration6.2 KB
language_model/attrs.jsonConfiguration78 B
preprocessor_config.jsonConfiguration262 B
special_tokens_map.jsonConfiguration85 B
README.mdDocumentation6.9 KB
full_eval.shOther1.4 KB
language_model/lm.binaryOther1.5 GB 01a8fba99ad0
language_model/unigrams.txtOther32.8 MB e7c72ae916af
log_mozilla-foundation_common_voice_6_0_ru_test_predictions.txtOther1.1 MB
log_mozilla-foundation_common_voice_6_0_ru_test_predictions_greedy.txtOther1.1 MB
log_mozilla-foundation_common_voice_6_0_ru_test_targets.txtOther1.1 MB
log_speech-recognition-community-v2_dev_data_ru_validation_predictions.txtOther211.9 KB
log_speech-recognition-community-v2_dev_data_ru_validation_predictions_greedy.txtOther212.3 KB
log_speech-recognition-community-v2_dev_data_ru_validation_targets.txtOther206.0 KB
mozilla-foundation_common_voice_6_0_ru_test_eval_results.txtOther50 B
mozilla-foundation_common_voice_6_0_ru_test_eval_results_greedy.txtOther49 B
speech-recognition-community-v2_dev_data_ru_validation_eval_results.txtOther48 B
speech-recognition-community-v2_dev_data_ru_validation_eval_results_greedy.txtOther48 B
.gitattributesRepository736 B
vocab.jsonTokenizer387 B

License and Download

License
apache-2.0
Access
Open weights, no gate
Download size
2.5 GB
Download from Jonatas Grosman

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

Built From

  • Trained on (disclosed) common_voice
  • Trained on (disclosed) mozilla-foundation/common_voice_6_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 ru Task Automatic Speech RecognitionMetric Test CERComparison conditions not established 2.88 jonatasgrosman
Publisher reported
Evaluated revision not stated
Common Voice ru Task Automatic Speech RecognitionMetric Test CER (+LM)Comparison conditions not established 2.24 jonatasgrosman
Publisher reported
Evaluated revision not stated
Common Voice ru Task Automatic Speech RecognitionMetric Test WERComparison conditions not established 13.3 jonatasgrosman
Publisher reported
Evaluated revision not stated
Common Voice ru Task Automatic Speech RecognitionMetric Test WER (+LM)Comparison conditions not established 9.57 jonatasgrosman
Publisher reported
Evaluated revision not stated
Robust Speech Event - Dev Data Task Automatic Speech RecognitionMetric Dev CERComparison conditions not established 14.8 jonatasgrosman
Publisher reported
Evaluated revision not stated
Robust Speech Event - Dev Data Task Automatic Speech RecognitionMetric Dev CER (+LM)Comparison conditions not established 13.5 jonatasgrosman
Publisher reported
Evaluated revision not stated
Robust Speech Event - Dev Data Task Automatic Speech RecognitionMetric Dev WERComparison conditions not established 40.22 jonatasgrosman
Publisher reported
Evaluated revision not stated
Robust Speech Event - Dev Data Task Automatic Speech RecognitionMetric Dev WER (+LM)Comparison conditions not established 33.61 jonatasgrosman
Publisher reported
Evaluated revision not stated

Memory Requirements

PrecisionWeights in memory
As published2.5 GB

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

Built on This Model

Questions About wav2vec2-large-xlsr-53-russian

Can I use wav2vec2-large-xlsr-53-russian commercially?

Yes. wav2vec2-large-xlsr-53-russian 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

wav2vec2-large-xlsr-53-japanese

Jonatas Grosman

Fine-tuned facebook/wav2vec2-large-xlsr-53 on Japanese using the train and validation splits of Common Voice 6.1, CSS10 and JSUT. When using this model, make sure that your speech input is sampled at 16kHz. This model has been fine-tuned thanks to the GPU credits generously given by the OVHcloud:) The script used for training can be found here: https://github.com/jonatasgrosman/wav2vec2-sprint The model can be used directly (without a language model) as follows... Using the HuggingSound library: The model can be evaluated as follows on the Japanese test data of Common Voice. In the table below I report the Word Error Rate (WER) and the Character Error Rate (CER) of the model. I ran the…

Open weights apache-2.0 transformers

Model · Speech recognition

whisperkit-coreml

Argmax

WhisperKit is part of Argmax OSS, an On-device Speech AI SDK for Apple Silicon: https://github.com/argmaxinc/argmax-oss-swift Check out the WhisperKit paper and presentation from ICML 2025: https://icml.cc/virtual/2025/47854 For real-time transcription with speakers and custom vocabulary, check out Argmax Pro SDK: https://www.argmaxinc.com/blog/argmax-sdk-2

Open weights mit whisperkit

Model · Speech recognition

speaker-diarization-3.1

Pyannote

Using this open-source model in production? Consider switching to pyannoteAI for better and faster options. This pipeline is the same as pyannote/speaker-diarization-3.0 except it removes the problematic use of onnxruntime. Both speaker segmentation and embedding now run in pure PyTorch. This should ease deployment and possibly speed up inference. It requires pyannote.audio version 3.1 or higher. It ingests mono audio sampled at 16kHz and outputs speaker diarization as an Annotation instance: - stereo or multi-channel audio files are automatically downmixed to mono by averaging the channels. - audio files sampled at a different rate are resampled to 16kHz automatically upon loading. 1.…

Access requested at publisher mit pyannote-audio

Fine-tuned facebook/wav2vec2-large-xlsr-53 on Portuguese using the train and validation splits of Common Voice 6.1. When using this model, make sure that your speech input is sampled at 16kHz. This model has been fine-tuned thanks to the GPU credits generously given by the OVHcloud:) The script used for training can be found here: https://github.com/jonatasgrosman/wav2vec2-sprint The model can be used directly (without a language model) as follows... Using the HuggingSound library: 1. To evaluate on mozilla-foundation/commonvoice60 with split test 2. To evaluate on speech-recognition-community-v2/devdata If you want to cite this model you can use this

Open weights apache-2.0 transformers

Model · Speech recognition

speaker-diarization-community-1

Pyannote

This pipeline ingests mono audio sampled at 16kHz and outputs speaker diarization. - stereo or multi-channel audio files are automatically downmixed to mono by averaging the channels. - audio files sampled at a different rate are resampled to 16kHz automatically upon loading. The main improvements brought by Community-1 are: - improved speaker assignment and counting - simpler reconciliation with transcription timestamps with exclusive speaker diarization - easy offline use (i.e. without internet connection) - (optionally) hosted on pyannoteAI cloud 1. pip install pyannote.audio 3. Create access token at hf.co/settings/tokens. Out of the box, Community-1 is much better than…

Access requested at publisher cc-by-4.0 pyannote-audio

Model · Speech recognition

wav2vec2-large-xlsr-53-polish

Jonatas Grosman

Fine-tuned facebook/wav2vec2-large-xlsr-53 on Polish using the train and validation splits of Common Voice 6.1. When using this model, make sure that your speech input is sampled at 16kHz. This model has been fine-tuned thanks to the GPU credits generously given by the OVHcloud:) The script used for training can be found here: https://github.com/jonatasgrosman/wav2vec2-sprint The model can be used directly (without a language model) as follows... Using the HuggingSound library: 1. To evaluate on mozilla-foundation/commonvoice60 with split test 2. To evaluate on speech-recognition-community-v2/devdata If you want to cite this model you can use this

Open weights apache-2.0 transformers