andrewbawitlung/wav2vec2-xls-r-1b-mizo-lus-v25

Mizo Automatic Speech Recognition

This model is a fine-tuned version of facebook/wav2vec2-xls-r-1b on the generator dataset. It achieves the following results on the evaluation set:

• Loss: 0.1267
• Wer: 0.1435

Citation

BibTeX entry and citation info:

@article{10.1145/3746063,
author = {Bawitlung, Andrew and Dash, Sandeep Kumar and Pattanayak, Radha Mohan},
title = {Mizo Automatic Speech Recognition: Leveraging Wav2vec 2.0 and XLS-R for Enhanced Accuracy in Low-Resource Language Processing},
year = {2025},
issue_date = {July 2025},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
volume = {24},
number = {7},
issn = {2375-4699},
url = {https://doi.org/10.1145/3746063},
doi = {10.1145/3746063},
journal = {ACM Trans. Asian Low-Resour. Lang. Inf. Process.},
month = jul,
articleno = {72},
numpages = {15},
}

Training and evaluation data

MiZonal v1.0

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.0003
• train_batch_size: 8
• eval_batch_size: 8
• seed: 50
• gradient_accumulation_steps: 8
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 1000
• num_epochs: 28
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
2.2954	2.18	300	0.3737	0.4528
0.6507	4.35	600	0.1903	0.2866
0.492	6.53	900	0.1740	0.2419
0.4302	8.7	1200	0.1503	0.2189
0.3512	10.88	1500	0.1344	0.1884
0.2963	13.06	1800	0.1264	0.2071
0.2536	15.23	2100	0.1250	0.1868
0.2075	17.41	2400	0.1217	0.1599
0.1775	19.58	2700	0.1121	0.1602
0.151	21.76	3000	0.1204	0.1601
0.1253	23.93	3300	0.1211	0.1435
0.1073	26.11	3600	0.1267	0.1521

Framework versions

• Transformers 4.37.2
• Pytorch 2.3.1+cu121
• Datasets 2.16.1
• Tokenizers 0.15.1