Target Speaker ASR with Whisper
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2024
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866915104924106752 |
|---|---|
| author | Polok, Alexander Klement, Dominik Wiesner, Matthew Khudanpur, Sanjeev Černocký, Jan Burget, Lukáš |
| author_facet | Polok, Alexander Klement, Dominik Wiesner, Matthew Khudanpur, Sanjeev Černocký, Jan Burget, Lukáš |
| contents | We propose a novel approach to enable the use of large, single-speaker ASR models, such as Whisper, for target speaker ASR. The key claim of this method is that it is much easier to model relative differences among speakers by learning to condition on frame-level diarization outputs than to learn the space of all speaker embeddings. We find that adding even a single bias term per diarization output type before the first transformer block can transform single-speaker ASR models into target-speaker ASR models. Our approach also supports speaker-attributed ASR by sequentially generating transcripts for each speaker in a diarization output. This simplified method outperforms baseline speech separation and diarization cascade by 12.9 % absolute ORC-WER on the NOTSOFAR-1 dataset. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2409_09543 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Target Speaker ASR with Whisper Polok, Alexander Klement, Dominik Wiesner, Matthew Khudanpur, Sanjeev Černocký, Jan Burget, Lukáš Audio and Speech Processing Sound We propose a novel approach to enable the use of large, single-speaker ASR models, such as Whisper, for target speaker ASR. The key claim of this method is that it is much easier to model relative differences among speakers by learning to condition on frame-level diarization outputs than to learn the space of all speaker embeddings. We find that adding even a single bias term per diarization output type before the first transformer block can transform single-speaker ASR models into target-speaker ASR models. Our approach also supports speaker-attributed ASR by sequentially generating transcripts for each speaker in a diarization output. This simplified method outperforms baseline speech separation and diarization cascade by 12.9 % absolute ORC-WER on the NOTSOFAR-1 dataset. |
| title | Target Speaker ASR with Whisper |
| topic | Audio and Speech Processing Sound |
| url | https://arxiv.org/abs/2409.09543 |