Wav2vec2 Large 10min Lv60 Self

🚀 Wav2Vec2-Large-10min-Lv60 + Self-Training

This project is a direct state_dict transfer from fairseq to huggingface, with identical weights. It is based on Facebook's Wav2Vec2. The large model is pretrained and fine - tuned on 10 minutes of Libri - Light and Librispeech with 16kHz sampled speech audio. It was trained with the Self - Training objective. Ensure your speech input is sampled at 16kHz when using the model.

🚀 Quick Start

Model Information

Authors

Alexei Baevski, Henry Zhou, Abdelrahman Mohamed, Michael Auli

Paper

Paper

Abstract

They show for the first time that learning powerful representations from speech audio alone followed by fine - tuning on transcribed speech can outperform the best semi - supervised methods while being conceptually simpler. wav2vec 2.0 masks the speech input in the latent space and solves a contrastive task defined over a quantization of the latent representations which are jointly learned. Experiments using all labeled data of Librispeech achieve 1.8/3.3 WER on the clean/other test sets. When lowering the amount of labeled data to one hour, wav2vec 2.0 outperforms the previous state of the art on the 100 hour subset while using 100 times less labeled data. Using just ten minutes of labeled data and pre - training on 53k hours of unlabeled data still achieves 4.8/8.2 WER. This demonstrates the feasibility of speech recognition with limited amounts of labeled data.

The original model can be found under https://github.com/pytorch/fairseq/tree/master/examples/wav2vec#wav2vec - 20.

💻 Usage Examples

Basic Usage

To transcribe audio files, the model can be used as a standalone acoustic model as follows:

from transformers import Wav2Vec2Processor, Wav2Vec2ForCTC
from datasets import load_dataset
import torch

# load model and processor
processor = Wav2Vec2Processor.from_pretrained("Splend1dchan/wav2vec2-large-10min-lv60-self")
model = Wav2Vec2ForCTC.from_pretrained("Splend1dchan/wav2vec2-large-10min-lv60-self")

# 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

# retrieve logits
logits = model(input_values).logits

# take argmax and decode
predicted_ids = torch.argmax(logits, dim=-1)
transcription = processor.batch_decode(predicted_ids)

Advanced Usage

This code snippet shows how to evaluate facebook's Splend1dchan/wav2vec2-large-10min-lv60-self on LibriSpeech's "clean" and "other" test data.

from datasets import load_dataset
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
import torch
from jiwer import wer
librispeech_eval = load_dataset("librispeech_asr", "clean", split="test")
model = Wav2Vec2ForCTC.from_pretrained("Splend1dchan/wav2vec2-large-10min-lv60-self").to("cuda")
processor = Wav2Vec2Processor.from_pretrained("Splend1dchan/wav2vec2-large-10min-lv60-self")
def map_to_pred(batch):
    inputs = processor(batch["audio"]["array"], return_tensors="pt", padding="longest")
    input_values = inputs.input_values.to("cuda")
    attention_mask = inputs.attention_mask.to("cuda")

    with torch.no_grad():
        logits = model(input_values, attention_mask=attention_mask).logits
    predicted_ids = torch.argmax(logits, dim=-1)
    transcription = processor.batch_decode(predicted_ids)
    batch["transcription"] = transcription
    return batch
result = librispeech_eval.map(map_to_pred, remove_columns=["speech"])
print("WER:", wer(result["text"], result["transcription"]))