TOGGL: Transcribing Overlapping Speech with Staggered Labeling
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866913466730676224 |
|---|---|
| author | Li, Chak-Fai Hartmann, William Snover, Matthew |
| author_facet | Li, Chak-Fai Hartmann, William Snover, Matthew |
| contents | Transcribing the speech of multiple overlapping speakers typically requires separating the audio into multiple streams and recognizing each one independently. More recent work jointly separates and transcribes, but requires a separate decoding component for each speaker. We propose the TOGGL model to simultaneously transcribe the speech of multiple speakers. The TOGGL model uses special output tokens to attribute the speech to each speaker with only a single decoder. Our approach generalizes beyond two speakers, even when trained only on two-speaker data. We demonstrate superior performance compared to competing approaches on a conversational speech dataset. Our approach also improves performance on single-speaker audio. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2408_06474 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | TOGGL: Transcribing Overlapping Speech with Staggered Labeling Li, Chak-Fai Hartmann, William Snover, Matthew Computation and Language Sound Audio and Speech Processing Transcribing the speech of multiple overlapping speakers typically requires separating the audio into multiple streams and recognizing each one independently. More recent work jointly separates and transcribes, but requires a separate decoding component for each speaker. We propose the TOGGL model to simultaneously transcribe the speech of multiple speakers. The TOGGL model uses special output tokens to attribute the speech to each speaker with only a single decoder. Our approach generalizes beyond two speakers, even when trained only on two-speaker data. We demonstrate superior performance compared to competing approaches on a conversational speech dataset. Our approach also improves performance on single-speaker audio. |
| title | TOGGL: Transcribing Overlapping Speech with Staggered Labeling |
| topic | Computation and Language Sound Audio and Speech Processing |
| url | https://arxiv.org/abs/2408.06474 |