Xls R 300m Es

🚀 Wav2Vec2-XLSR-300m-es

This model is a fine - tuned version of facebook/wav2vec2-xls-r-300m on the Spanish Common Voice dataset. Thanks to the GPU credits generously provided by OVHcloud for the Speech Recognition challenge, it achieves excellent performance in speech recognition tasks.

✨ Features

• Fine - tuned on Spanish Data: Trained on the Spanish Common Voice dataset, making it highly suitable for Spanish speech recognition.
• Support for n - gram: Can be used with a 5 - gram model included in the processor, improving recognition accuracy.
• Good Evaluation Results: Achieves low WER (Word Error Rate) on multiple evaluation datasets.

📦 Installation

No specific installation steps are provided in the original document, so this section is skipped.

💻 Usage Examples

Basic Usage

The model can be used with n - gram (n = 5) included in the processor as follows.

import re
from transformers import AutoModelForCTC,Wav2Vec2ProcessorWithLM
import torch

# Loading model and processor
processor = Wav2Vec2ProcessorWithLM.from_pretrained("polodealvarado/xls-r-300m-es")
model = AutoModelForCTC.from_pretrained("polodealvarado/xls-r-300m-es")

# Cleaning characters
def remove_extra_chars(batch):
    chars_to_ignore_regex = '[^a-záéíóúñ ]'
    text = batch["translation"][target_lang]
    batch["text"] = re.sub(chars_to_ignore_regex, "", text.lower())
    return batch
    
# Preparing dataset
def prepare_dataset(batch):
    audio = batch["audio"]
    batch["input_values"] = processor(audio["array"], sampling_rate=audio["sampling_rate"],return_tensors="pt",padding=True).input_values[0]    
    with processor.as_target_processor():
        batch["labels"] = processor(batch["sentence"]).input_ids
    return batch
  

common_voice_test = load_dataset("mozilla-foundation/common_voice_8_0", "es", split="test",use_auth_token=True)
common_voice_test = common_voice_test.remove_columns(["accent", "age", "client_id", "down_votes", "gender", "locale", "segment", "up_votes"])
common_voice_test = common_voice_test.cast_column("audio", Audio(sampling_rate=16_000))        
common_voice_test = common_voice_test.map(remove_extra_chars, remove_columns=dataset.column_names)
common_voice_test = common_voice_test.map(prepare_dataset)

# Testing first sample
inputs = torch_tensor(common_voice_test[0]["input_values"])

with torch.no_grad():
    logits = model(inputs).logits

pred_ids = torch.argmax(logits, dim=-1)
text = processor.batch_decode(logits.numpy()).text
print(text) # 'bien y qué regalo vas a abrir primero'

Advanced Usage

On the other hand, you can execute the eval.py file for evaluation.

# To use GPU: --device 0
$ python eval.py --model_id polodealvarado/xls-r-300m-es --dataset mozilla-foundation/common_voice_8_0 --config es --device 0 --split test

📚 Documentation

Evaluation Results

It achieves the following results on the evaluation set: Without LM:

• Loss : 0.1900
• Wer : 0.146

With 5 - gram:

• WER: 0.109
• CER: 0.036

Model Index

Training Hyperparameters

The following hyperparameters were used during training:

Training Results

Framework Versions

🔧 Technical Details

No specific technical details beyond what is already covered in the documentation section are provided, so this section is skipped.

📄 License

The model is released under the Apache - 2.0 license.