karolnowakowski/wav2vec2-large-xlsr-53-ain-sakh

Wav2Vec2-Large-XLSR-53 fine-tuned for automatic transcription of Sakhalin Ainu

This is a wav2vec-large-xlsr-53 model after continued pretraining on speech data in Hokkaido Ainu and Sakhalin Ainu (see wav2vec2-large-xlsr-53-pretrain-ain) and fine-tuning for automatic speech recognition on 10h of labeled Sakhalin Ainu data. For details, please refer to the paper.

On our evaluation set, the model yielded the following results:

• CER: 9.7
• WER: 29.8

Citation

When using the model please cite the following paper:

@article{NOWAKOWSKI2023103148,
title = {Adapting multilingual speech representation model for a new, underresourced language through multilingual fine-tuning and continued pretraining},
journal = {Information Processing & Management},
volume = {60},
number = {2},
pages = {103148},
year = {2023},
issn = {0306-4573},
doi = {https://doi.org/10.1016/j.ipm.2022.103148},
url = {https://www.sciencedirect.com/science/article/pii/S0306457322002497},
author = {Karol Nowakowski and Michal Ptaszynski and Kyoko Murasaki and Jagna Nieuważny},
keywords = {Automatic speech transcription, ASR, Wav2vec 2.0, Pretrained transformer models, Speech representation models, Cross-lingual transfer, Language documentation, Endangered languages, Underresourced languages, Sakhalin Ainu},
abstract = {In recent years, neural models learned through self-supervised pretraining on large scale multilingual text or speech data have exhibited promising results for underresourced languages, especially when a relatively large amount of data from related language(s) is available. While the technology has a potential for facilitating tasks carried out in language documentation projects, such as speech transcription, pretraining a multilingual model from scratch for every new language would be highly impractical. We investigate the possibility for adapting an existing multilingual wav2vec 2.0 model for a new language, focusing on actual fieldwork data from a critically endangered tongue: Ainu. Specifically, we (i) examine the feasibility of leveraging data from similar languages also in fine-tuning; (ii) verify whether the model’s performance can be improved by further pretraining on target language data. Our results show that continued pretraining is the most effective method to adapt a wav2vec 2.0 model for a new language and leads to considerable reduction in error rates. Furthermore, we find that if a model pretrained on a related speech variety or an unrelated language with similar phonological characteristics is available, multilingual fine-tuning using additional data from that language can have positive impact on speech recognition performance when there is very little labeled data in the target language.}
}