full record print

Design choices for PixIT-based speaker-attributed ASR: Team ToTaTo at the NOTSOFAR-1 challenge

author
statement of authorship
Joonas Kalda, Séverin Baroudi, Martin Lebourdais, Clément Pagés, Ricard Marxer, Tanel Alumäe, Hervé Bredin
source
Computer Speech & Language
publisher
journal volume number month
vol. 95
year of publication
pages
art. 101824, 16 p. : ill
url
ISSN
0885-2308
1095-8363
notes
Bibliogr. p. 15-16
Open Access
Open Access
scientific publication
teaduspublikatsioon
language
inglise
subject term
keyword
Speaker diarization
Speaker-attributed automatic speech
recognition
Speaker embeddings
SSL models
Joint training
Kalda, J., Baroudi, S., Lebourdais, M., Pagés, C., Marxer, R., Alumäe, T., Bredin, H. Design choices for PixIT-based speaker-attributed ASR: Team ToTaTo at the NOTSOFAR-1 challenge // Computer Speech & Language (2026) vol. 95, art. 101824, 16 p. : ill. https://doi.org/10.1016/j.csl.2025.101824