WAV2VEC2-DUTCH-LARGE

Introduction

The WAV2VEC2-DUTCH-LARGE model is a Dutch language model based on Wav2Vec2. It is developed by further pre-training the original English facebook/wav2vec2-large model on Dutch speech data sourced from Het Corpus Gesproken Nederlands.

Architecture

This model utilizes the Wav2Vec2 architecture, which is designed for speech-related tasks. It builds on the pre-existing English model by adapting it to the Dutch language through additional training on a relevant Dutch speech corpus.

Training

The original facebook/wav2vec2-large model was pre-trained on English data. For the Dutch version, the model underwent further pre-training using the Het Corpus Gesproken Nederlands dataset. This dataset is specifically curated to enhance the model's performance in recognizing and processing Dutch speech.

Guide: Running Locally

1. Setup Environment: Ensure you have Python and PyTorch installed. Use pip to install the transformers library from Hugging Face.

   pip install transformers
   

2. Load the Model: Use the Hugging Face transformers library to load the model.

   from transformers import Wav2Vec2ForCTC, Wav2Vec2Tokenizer
   
   tokenizer = Wav2Vec2Tokenizer.from_pretrained("GroNLP/wav2vec2-dutch-large")
   model = Wav2Vec2ForCTC.from_pretrained("GroNLP/wav2vec2-dutch-large")
   

3. Inference: Prepare your Dutch audio input, tokenize it, and use the model to perform inference.

4. Hardware: For optimal performance, especially with large models, consider using cloud GPUs such as those provided by AWS, Google Cloud, or Azure.

License

The model is available under the terms specified on its Hugging Face model card. Ensure compliance with these terms when using it in your applications.