Wav2vec2 Large Xlsr Galician

🚀 wav2vec2-large-xlsr-galician

A fine - tuned model for Galician language based on self - supervised learning, enabling high - quality automatic speech recognition.

🚀 Quick Start

The wav2vec2-large-xlsr-galician is a fine - tuned model for Galician language. It is based on the facebook/wav2vec2-large-xlsr-53 self - supervised model and fine - tuned with audio labelled from OpenSLR and Mozilla Common_Voice (both datasets previously refined).

✨ Features

• Language Specific: Specifically fine - tuned for the Galician language.
• Based on Self - Supervised Learning: Built upon the powerful wav2vec2-large-xlsr-53 self - supervised model.
• Trained on Multiple Datasets: Utilizes data from OpenSLR and Mozilla Common Voice.

📦 Installation

No specific installation steps are provided in the original document.

💻 Usage Examples

Basic Usage

from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor

model = Wav2Vec2ForCTC.from_pretrained("ifrz/wav2vec2-large-xlsr-galician")
processor = Wav2Vec2Processor.from_pretrained("ifrz/wav2vec2-large-xlsr-galician")

# Reading taken audio clip
import librosa, torch
audio, rate = librosa.load("./gl_test_1.wav", sr = 16000)

# Taking an input value
input_values = processor(audio, sampling_rate=16_000, return_tensors = "pt", padding="longest").input_values
# Storing logits (non-normalized prediction values)
logits = model(input_values).logits
# Storing predicted ids
prediction = torch.argmax(logits, dim = -1)

# Passing the prediction to the tokenzer decode to get the transcription
transcription = processor.batch_decode(prediction)[0]
print(transcription)

Advanced Usage

Before using the model, make sure that the audio speech input is sampled at 16kHz (mono).

📚 Documentation

Model Information

Model Index

• Name: Galician wav2vec2 - large - xlsr - galician
• Results:
  • Task:
    • Name: Speech Recognition
    • Type: automatic - speech - recognition
  • Datasets:
    • Dataset 1:
      • Name: OpenSLR
      • Type: openslr
      • Args: gl
    • Dataset 2:
      • Name: mozilla - foundation
      • Type: common voice
      • Args: gl
  • Metrics:
    • Name: Test WER
    • Type: wer
    • Value: 7.12

📄 License

This project is licensed under the apache - 2.0 license.