Guardado en:
| Autores principales: | , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | https://arxiv.org/abs/2505.17088 |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866917284615815168 |
|---|---|
| author | Attia, Ahmed Adel Demszky, Dorottya Liu, Jing Espy-Wilson, Carol |
| author_facet | Attia, Ahmed Adel Demszky, Dorottya Liu, Jing Espy-Wilson, Carol |
| contents | Recent progress in speech recognition has relied on models trained on vast amounts of labeled data. However, classroom Automatic Speech Recognition (ASR) faces the real-world challenge of abundant weak transcripts paired with only a small amount of accurate, gold-standard data. In such low-resource settings, high transcription costs make re-transcription impractical. To address this, we ask: what is the best approach when abundant inexpensive weak transcripts coexist with limited gold-standard data, as is the case for classroom speech data? We propose Weakly Supervised Pretraining (WSP), a two-step process where models are first pretrained on weak transcripts in a supervised manner, and then fine-tuned on accurate data. Our results, based on both synthetic and real weak transcripts, show that WSP outperforms alternative methods, establishing it as an effective training methodology for low-resource ASR in real-world scenarios. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2505_17088 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | From Weak Labels to Strong Results: Utilizing 5,000 Hours of Noisy Classroom Transcripts with Minimal Accurate Data Attia, Ahmed Adel Demszky, Dorottya Liu, Jing Espy-Wilson, Carol Audio and Speech Processing Computation and Language Sound Recent progress in speech recognition has relied on models trained on vast amounts of labeled data. However, classroom Automatic Speech Recognition (ASR) faces the real-world challenge of abundant weak transcripts paired with only a small amount of accurate, gold-standard data. In such low-resource settings, high transcription costs make re-transcription impractical. To address this, we ask: what is the best approach when abundant inexpensive weak transcripts coexist with limited gold-standard data, as is the case for classroom speech data? We propose Weakly Supervised Pretraining (WSP), a two-step process where models are first pretrained on weak transcripts in a supervised manner, and then fine-tuned on accurate data. Our results, based on both synthetic and real weak transcripts, show that WSP outperforms alternative methods, establishing it as an effective training methodology for low-resource ASR in real-world scenarios. |
| title | From Weak Labels to Strong Results: Utilizing 5,000 Hours of Noisy Classroom Transcripts with Minimal Accurate Data |
| topic | Audio and Speech Processing Computation and Language Sound |
| url | https://arxiv.org/abs/2505.17088 |