Data2vec Audio Large 100h

🚀 Data2Vec-Audio-Large-100h

A large model pre - trained and fine - tuned on 100 hours of Librispeech audio sampled at 16kHz.

🚀 Quick Start

This is a large model that has been pre - trained and fine - tuned on 100 hours of Librispeech audio sampled at 16kHz. When using the model, ensure that your speech input is also sampled at 16kHz.

✨ Features

• General Self - Supervised Learning: Based on Facebook's Data2Vec, it uses the same learning method for speech, NLP, or computer vision.
• State - of - the - Art Performance: Experiments on major benchmarks of speech recognition, image classification, and natural language understanding show state - of - the - art or competitive performance.

📚 Documentation

Paper

Paper

Authors

Alexei Baevski, Wei - Ning Hsu, Qiantong Xu, Arun Babu, Jiatao Gu, Michael Auli

Abstract

While the general idea of self - supervised learning is identical across modalities, the actual algorithms and objectives differ widely because they were developed with a single modality in mind. To get us closer to general self - supervised learning, we present data2vec, a framework that uses the same learning method for either speech, NLP or computer vision. The core idea is to predict latent representations of the full input data based on a masked view of the input in a self - distillation setup using a standard Transformer architecture. Instead of predicting modality - specific targets such as words, visual tokens or units of human speech which are local in nature, data2vec predicts contextualized latent representations that contain information from the entire input. Experiments on the major benchmarks of speech recognition, image classification, and natural language understanding demonstrate a new state of the art or competitive performance to predominant approaches.

Pre - Training method

For more information, please take a look at the official paper.

💻 Usage Examples

Basic Usage

 from transformers import Wav2Vec2Processor, Data2VecForCTC
 from datasets import load_dataset
 import torch
 
 # load model and processor
 processor = Wav2Vec2Processor.from_pretrained("facebook/data2vec-audio-large-100h")
 model = Data2VecForCTC.from_pretrained("facebook/data2vec-audio-large-100h")
     
 # load dummy dataset and read soundfiles
 ds = load_dataset("patrickvonplaten/librispeech_asr_dummy", "clean", split="validation")
 
 # tokenize
 input_values = processor(ds[0]["audio"]["array"], return_tensors="pt", padding="longest").input_values  # Batch size 1
 
 # retrieve logits
 logits = model(input_values).logits
 
 # take argmax and decode
 predicted_ids = torch.argmax(logits, dim=-1)
 transcription = processor.batch_decode(predicted_ids)

📄 License

This project is licensed under the Apache - 2.0 license.

📦 Additional Information