This model is a finetuned version of facebook/wav2vec2-large-xlsr-53. There was an issue with vocab, seems like there are special characters included and they were not considered during training You could try
Open-weight model · Speech recognition
Wav2Vec2-large-xlsr-hindi
by Shyam Sunder Kumar theainerd/Wav2Vec2-large-xlsr-hindi
Fine-tuned facebook/wav2vec2-large-xlsr-53 hindi using the Multilingual and code-switching ASR challenges for low resource Indian languages. When using this model, make sure that your speech input is sampled at 16kHz.
Parameters316M
Context—
Weights2.5 GB
License—
AccessOpen weights
Monthly Downloads1.4M
Runs On
What it takes to serve Wav2Vec2-large-xlsr-hindi (316M parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them.
| Precision | Weights | Memory needed | Cheapest setup | Per hour | Also 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-large-xlsr-hindi
Feed it 16 kHz audio; the publisher is explicit about the sample rate. It is a Hindi speech recognizer, 316 million parameters in a Wav2Vec2ForCTC architecture with 24 layers and a 64-entry vocabulary, fine-tuned by Shyam Sunder Kumar from facebook/wav2vec2-large-xlsr-53 for low resource Indian languages. Memory is not the constraint: 0.8 GB at 16-bit, 0.4 GB at 8-bit, 0.2 GB at 4-bit. Our cheapest priced card, a 192 GB MI300X at $1.85 an hour, holds it hundreds of times over, so the deployment shape is many audio streams sharing one device.
The license field is empty. Open access is not a grant of commercial rights, so get the publisher's terms in writing before this goes into a product, and check the upstream base, facebook/wav2vec2-large-xlsr-53, since its terms bound what this derivative can do. It runs directly without a language model and was evaluated on the Hindi test data of Common Voice, so reproduce that result on your own audio before committing.
Model Card
Fine-tuned facebook/wav2vec2-large-xlsr-53 hindi using the Multilingual and code-switching ASR challenges for low resource Indian languages. 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 as follows on the hindi test data of Common Voice. The script used for training can be found Hindi ASR Fine Tuning Wav2Vec2
Excerpt from the card by Shyam Sunder Kumar.
Configuration
- Architecture
- Wav2Vec2ForCTC
- Layers
- 24
- Hidden size
- 1,024
- Feed-forward size
- 4,096
- Attention heads
- 16
- Vocabulary size
- 64
- Model type
- wav2vec2
Identity and Version
- Repository
- theainerd/Wav2Vec2-large-xlsr-hindi
- Publisher
- Shyam Sunder Kumar
- Task
- Speech recognition
- Modality
- Audio
- Library
- transformers
- Parameters
- 316M parameters
- Languages
- hi
- Revision
- 062f7f566e2671336992b011dcb9387cd3cffe5e
- First published
- 2022-03-02
- Last updated
- 2025-04-01
Files and Weights
9 files, 2.5 GB in total. The weights are 2 files totalling 2.5 GB in bin, safetensors.
Weights2 files · 2.5 GB
Configuration3 files · 1.8 KB
Tokenizer2 files · 834 B
Documentation1 file · 3.6 KB
Repository1 file · 744 B
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| model.safetensors | Weights | 1.3 GB | c9469047fff5 |
| pytorch_model.bin | Weights | 1.3 GB | 7af3f44f6dd0 |
| config.json | Configuration | 1.6 KB | — |
| preprocessor_config.json | Configuration | 158 B | — |
| special_tokens_map.json | Configuration | 85 B | — |
| README.md | Documentation | 3.6 KB | — |
| .gitattributes | Repository | 744 B | — |
| tokenizer_config.json | Tokenizer | 138 B | — |
| vocab.json | Tokenizer | 696 B | — |
License and Download
- License
- Not stated by the source
- Access
- Open weights, no gate
- Download size
- 2.5 GB
Download from Shyam Sunder Kumar
Released by Shyam Sunder Kumar through its official repository on Hugging Face.
Built From
- Derived from facebook/wav2vec2-large-xlsr-53
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 2.5 GB |
| 16-bit | 0.6 GB |
| 8-bit | 0.3 GB |
| 4-bit | 0.2 GB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.
Questions About Wav2Vec2-large-xlsr-hindi
How much GPU memory does Wav2Vec2-large-xlsr-hindi need?
About 0.8 GB at 16-bit and 0.2 GB at 4-bit: the weights (316M parameters) plus a working margin. A long context needs more.
What is the cheapest GPU to run Wav2Vec2-large-xlsr-hindi 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.
Similar Models
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…
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 %…
Model · Speech recognition
wav2vec2-xls-r-parlaspeech-hr
CLASSLA - CLARIN Knowledge Centre for South Slavic Languages
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…
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…
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…