DNCASR: End-to-End Training for Speaker-Attributed ASR
Fuente:
arXiv
Guardado en:
| Autores principales: | , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866909633037205504 |
|---|---|
| author | Zheng, Xianrui Zhang, Chao Woodland, Philip C. |
| author_facet | Zheng, Xianrui Zhang, Chao Woodland, Philip C. |
| contents | This paper introduces DNCASR, a novel end-to-end trainable system designed for joint neural speaker clustering and automatic speech recognition (ASR), enabling speaker-attributed transcription of long multi-party meetings. DNCASR uses two separate encoders to independently encode global speaker characteristics and local waveform information, along with two linked decoders to generate speaker-attributed transcriptions. The use of linked decoders allows the entire system to be jointly trained under a unified loss function. By employing a serialised training approach, DNCASR effectively addresses overlapping speech in real-world meetings, where the link improves the prediction of speaker indices in overlapping segments. Experiments on the AMI-MDM meeting corpus demonstrate that the jointly trained DNCASR outperforms a parallel system that does not have links between the speaker and ASR decoders. Using cpWER to measure the speaker-attributed word error rate, DNCASR achieves a 9.0% relative reduction on the AMI-MDM Eval set. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2506_01916 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | DNCASR: End-to-End Training for Speaker-Attributed ASR Zheng, Xianrui Zhang, Chao Woodland, Philip C. Audio and Speech Processing This paper introduces DNCASR, a novel end-to-end trainable system designed for joint neural speaker clustering and automatic speech recognition (ASR), enabling speaker-attributed transcription of long multi-party meetings. DNCASR uses two separate encoders to independently encode global speaker characteristics and local waveform information, along with two linked decoders to generate speaker-attributed transcriptions. The use of linked decoders allows the entire system to be jointly trained under a unified loss function. By employing a serialised training approach, DNCASR effectively addresses overlapping speech in real-world meetings, where the link improves the prediction of speaker indices in overlapping segments. Experiments on the AMI-MDM meeting corpus demonstrate that the jointly trained DNCASR outperforms a parallel system that does not have links between the speaker and ASR decoders. Using cpWER to measure the speaker-attributed word error rate, DNCASR achieves a 9.0% relative reduction on the AMI-MDM Eval set. |
| title | DNCASR: End-to-End Training for Speaker-Attributed ASR |
| topic | Audio and Speech Processing |
| url | https://arxiv.org/abs/2506.01916 |