distilxlsr_bp_16-24 / README.md
lgris's picture
Create README.md
61102aa
metadata
language: pt
tags:
  - speech
license: apache-2.0

DistilXLSR-53 for BP

DistilXLSR-53 for BP: DistilHuBERT applied to Wav2vec XLSR-53 for Brazilian Portuguese

The base model pretrained on 16kHz sampled speech audio. When using the model make sure that your speech input is also sampled at 16Khz.

Note: This model does not have a tokenizer as it was pretrained on audio alone. In order to use this model speech recognition, a tokenizer should be created and the model should be fine-tuned on labeled text data. Check out this blog for more in-detail explanation of how to fine-tune the model. Paper: DistilHuBERT: Speech Representation Learning by Layer-wise Distillation of Hidden-unit BERT Authors: Heng-Jui Chang, Shu-wen Yang, Hung-yi Lee

Note 2: The XLSR-53 model was distilled using Brazilian Portuguese Datasets for test purposes. The dataset is quite small to perform such task (the performance might not be so good as the original work).

Abstract

Self-supervised speech representation learning methods like wav2vec 2.0 and Hidden-unit BERT (HuBERT) leverage unlabeled speech data for pre-training and offer good representations for numerous speech processing tasks. Despite the success of these methods, they require large memory and high pre-training costs, making them inaccessible for researchers in academia and small companies. Therefore, this paper introduces DistilHuBERT, a novel multi-task learning framework to distill hidden representations from a HuBERT model directly. This method reduces HuBERT's size by 75% and 73% faster while retaining most performance in ten different tasks. Moreover, DistilHuBERT required little training time and data, opening the possibilities of pre-training personal and on-device SSL models for speech.

Usage

See this blog for more information on how to fine-tune the model.