MMS-300M fine-tuned on WAXAL — Wolaytta

This model is part of WAXALNet, a suite of ASR models fine-tuned on the WAXAL corpus across 19 African languages, developed as part of the WAXAL ASR Benchmark study.

Model Details


Language	Wolaytta (`wal`)
Language Family	Afro-Asiatic
Architecture	MMS-300M (300M parameters)
Base Model	facebook/mms-300m
Training Data	WAXAL corpus (conversational spontaneous speech)
Test WER	38.8%
Test CER	12.3%
License	cc-by-nc-4.0

Intended Use

This model is intended for automatic speech recognition of Wolaytta conversational speech. It was evaluated on the WAXAL test set (spontaneous, image-prompted speech) and partially on FLEURS (read speech). It is suitable for research and low-resource ASR applications. It is not recommended for high-stakes production use without further validation.

Training Data

Fine-tuned on the WAXAL corpus, a large-scale dataset of transcribed, image-prompted spontaneous speech across 19 African languages recorded in participants' natural environments. The Wolaytta training split contains conversational speech across diverse speakers. Data is released under CC-BY 4.0.

Usage

from transformers import pipeline

asr = pipeline("automatic-speech-recognition",
               model="waxal-benchmarking/mms-300m-waxal-wal")
result = asr("audio.wav")
print(result["text"])

Test Set Performance (WAXAL Benchmark)

Evaluated on the filtered WAXAL test set (duration >= 1.5s, speech rate >= 4 WPS).

Metric	Score
WER	38.8%
CER	12.3%

Full benchmark results across all 19 languages and 6 models are reported in the WAXAL ASR Benchmark paper (citation below).

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 0.0001
train_batch_size: 4
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 8
total_train_batch_size: 32
optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 500
num_epochs: 30
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
12.1194	0.3828	500	1.1588	0.9327	0.2974
3.3066	0.7656	1000	0.3395	0.4133	0.0837
3.4029	1.1478	1500	0.2819	0.3620	0.0724
2.8917	1.5305	2000	0.2671	0.3509	0.0702
2.8067	1.9133	2500	0.2596	0.3393	0.0673
2.7754	2.2955	3000	0.2539	0.3458	0.0684
2.7298	2.6783	3500	0.2498	0.3336	0.0664
2.9892	3.0605	4000	0.2433	0.3336	0.0661
2.7518	3.4433	4500	0.2403	0.3237	0.0639
2.7466	3.8260	5000	0.2421	0.3244	0.0645
2.6956	4.2082	5500	0.2416	0.3239	0.0638
2.7353	4.5910	6000	0.2412	0.3195	0.0634

Framework versions

Transformers 5.0.0
Pytorch 2.10.0+cu128
Datasets 4.0.0
Tokenizers 0.22.2

Citation

@article{waxalnet2026,
  title  = {The WAXAL ASR Benchmark: Fine-Tuned Edge Models Across 19 African Languages},
  author = {Olufemi, Victor Tolulope and Babatunde, Oreoluwa and Njema, Ramsey and
             Gbotemi, Bolarinwa and Yen, Wanchi Lucia and Uzodinma, John and
             Ajayi, Sunday and Williams, Oluwademilade and Moshood, Kausar and
             Anyaele, Innocent Elendu and Arefaine, Akebert Tesfahunegn and
             Hunzwi, Candace and Daniel, Wongel Dawit and Namuganga, Emmilly Immaculate and
             Kadima, Cleophas and Bahizire, Athanase Biluge and Ranaivoson, Onitsiky and
             Aaron, Emmanuel and Ladislaus, Nicholaus Dismas and Muhammed, Idris and
             Simenya, Jonathan Enoch and Koome, Martin and Endaylalu, Matewos Tegete and
             Adeyemo, Peter Ifeoluwa and Birindwa, Hondi Prisca and Eze-Mbey, Ukachi Agnes and
             Oduro-Yeboah, Yacoba and Aremu, Toluwani and Adjovi, Pericles and
             Ngueajio, Mikel K and Mitra, Prasenjit},
  year   = {2026},
  note   = {arXiv preprint arXiv:2606.02375}
}

Authors

Victor Tolulope Olufemi · Oreoluwa Babatunde · Ramsey Njema · Bolarinwa Gbotemi · Wanchi Lucia Yen · John Uzodinma · Sunday Ajayi · Oluwademilade Williams · Kausar Moshood · Innocent Elendu Anyaele · Akebert Tesfahunegn Arefaine · Candace Hunzwi · Wongel Dawit Daniel · Emmilly Immaculate Namuganga · Cleophas Kadima · Athanase Biluge Bahizire · Onitsiky Ranaivoson · Emmanuel Aaron · Nicholaus Dismas Ladislaus · Idris Muhammed · Jonathan Enoch Simenya · Martin Koome · Matewos Tegete Endaylalu · Peter Ifeoluwa Adeyemo · Hondi Prisca Birindwa · Ukachi Agnes Eze-Mbey · Yacoba Oduro-Yeboah · Toluwani Aremu · Pericles Adjovi · Mikel K Ngueajio · Prasenjit Mitra

Acknowledgements

We thank the following contributors for their language expertise and native-speaker evaluation support: Ajara Oyinloye, Abubakari Sadic Mohammed, Hafiz Adjei, Aliga Norah Lele, Marie-Louise B. Ndamuso, and Odong Diana.

This work was supported by Lynguallabs (compute, researchers & storage), Open Token (compute resources), and CMU Africa (researchers & native speakers).