Continued Pretraining for Low-Resource Swahili ASR: Achieving State-of-the-Art Performance with Minimal Labeled Data
Fuente:
arXiv
Salvato in:
| Autori principali: | , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866918383816015872 |
|---|---|
| author | Mutisya, Hillary Mugane, John |
| author_facet | Mutisya, Hillary Mugane, John |
| contents | We investigate continued pretraining (CPT) for adapting wav2vec2-bert-2.0 to Swahili automatic speech recognition (ASR). Our approach combines unlabeled audio with limited labeled data through pseudo-labeled CPT followed by supervised finetuning. With 20,000 labeled samples, we achieve 3.24% WER on Common Voice Swahili-an 82% relative improvement over the baseline. This result surpasses the best previously reported academic system (8.3% WER from XLS-R) by 61% relative improvement. We provide concrete data requirements and a replicable methodology applicable to other low-resource languages. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_11378 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Continued Pretraining for Low-Resource Swahili ASR: Achieving State-of-the-Art Performance with Minimal Labeled Data Mutisya, Hillary Mugane, John Sound Machine Learning Audio and Speech Processing We investigate continued pretraining (CPT) for adapting wav2vec2-bert-2.0 to Swahili automatic speech recognition (ASR). Our approach combines unlabeled audio with limited labeled data through pseudo-labeled CPT followed by supervised finetuning. With 20,000 labeled samples, we achieve 3.24% WER on Common Voice Swahili-an 82% relative improvement over the baseline. This result surpasses the best previously reported academic system (8.3% WER from XLS-R) by 61% relative improvement. We provide concrete data requirements and a replicable methodology applicable to other low-resource languages. |
| title | Continued Pretraining for Low-Resource Swahili ASR: Achieving State-of-the-Art Performance with Minimal Labeled Data |
| topic | Sound Machine Learning Audio and Speech Processing |
| url | https://arxiv.org/abs/2603.11378 |