CoolFace
Modelpublic

servinosmanov/whisper-large-v3-crh

sourceHugging Faceapache-2.0updated 9mo agoView on Hugging Face
0likes486downloads
Model Card

Whisper Large-v3 Fine-tuned for Crimean Tatar (Qırımtatarca)

This model is a fine-tuned version of openai/whisper-large-v3 for Automatic Speech Recognition (ASR) in Crimean Tatar language.

Model Description

  • —Base model: openai/whisper-large-v3 (1.54B parameters)
  • —Language: Crimean Tatar (crh) - both Cyrillic and Latin scripts
  • —Fine-tuning dataset: servinosmanov/tts-crh-sevil-fixed
  • —Best WER: 13.67% (with repetition penalty)

Training Details

  • —Epochs: ~17 (early stopping to prevent overfitting)
  • —Best checkpoint: epoch 12.7 with WER 16.0%
  • —Learning rate: 1e-5
  • —Batch size: 4 (with gradient accumulation steps: 4)
  • —Optimizer: AdamW

Usage

python
from transformers import WhisperForConditionalGeneration, WhisperProcessor
import torch
import librosa

# Load model
model_name = "servinosmanov/whisper-large-v3-crh"
processor = WhisperProcessor.from_pretrained(model_name)
model = WhisperForConditionalGeneration.from_pretrained(model_name)

device = "cuda" if torch.cuda.is_available() else "cpu"
model = model.to(device)
model.eval()

# Load and process audio
audio, sr = librosa.load("your_audio.wav", sr=16000)

input_features = processor(
    audio,
    sampling_rate=16000,
    return_tensors="pt"
).input_features.to(device)

# Generate transcription (with repetition penalty recommended)
with torch.no_grad():
    predicted_ids = model.generate(
        input_features,
        max_length=225,
        num_beams=5,
        repetition_penalty=1.2,
        no_repeat_ngram_size=3,
    )

transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)[0]
print(transcription)

Important Notes

  • —Repetition Penalty: It's recommended to use repetition_penalty=1.2 and no_repeat_ngram_size=3 during inference to prevent repetition loops
  • —Script Support: Works with both Cyrillic (Меним адым...) and Latin (Menim adım...) scripts

Comparison with whisper-medium

ModelWER
whisper-medium (fine-tuned)26.47%
whisper-large-v3 (this model)13.67%
Improvement48.4% relative

Limitations

  • —Trained primarily on female voice (Sevil speaker)
  • —May have reduced accuracy on very long utterances
  • —Best results with clear audio at 16kHz sampling rate

License

Apache 2.0 (same as base Whisper model)

Citation

If you use this model, please cite:

bibtex
@misc{whisper-large-v3-crh,
  author = {Servin Osmanov},
  title = {Whisper Large-v3 Fine-tuned for Crimean Tatar},
  year = {2025},
  publisher = {HuggingFace},
  url = {https://huggingface.co/servinosmanov/whisper-large-v3-crh}
}