whisper-large-v2-fa/README.md

library_name, language, license, base_model, tags, datasets, model-index

library_name	language	license	base_model	tags	datasets	model-index
transformers	fa	apache-2.0	openai/whisper-large-v2	generated_from_trainer	mozilla-foundation/common_voice_17_0	name results Whisper large-v2 Fa - Common Voice

Whisper large-v2 Fa - Common Voice

This model is a fine-tuned version of openai/whisper-large-v2 on the Common Voice 17.0 dataset.

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 1e-05
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 4000
• mixed_precision_training: Native AMP

Test results

• Best test WER (Word Error Rate): 0.322
• Best test CER (Character Error Rate): 0.106

Usage

import torch
from transformers import AutoModelForSpeechSeq2Seq, AutoProcessor, pipeline

device = "cuda:0" if torch.cuda.is_available() else "cpu"
torch_dtype = torch.float16 if torch.cuda.is_available() else torch.float32

model_id = "aictsharif/whisper-large-v2-fa"
model = AutoModelForSpeechSeq2Seq.from_pretrained(
    model_id, torch_dtype=torch_dtype, low_cpu_mem_usage=True, use_safetensors=True
)
model.to(device)

processor = AutoProcessor.from_pretrained(model_id)
pipe = pipeline(
    "automatic-speech-recognition",
    model=model,
    tokenizer=processor.tokenizer,
    feature_extractor=processor.feature_extractor,
    torch_dtype=torch_dtype,
    device=device,
)

result = pipe('sample.mp3')
print(result["text"])

Framework versions

• Transformers 4.51.3
• Pytorch 2.5.1+cu124
• Datasets 3.5.0
• Tokenizers 0.21.1