A per-head learned linear bias on relative positions lets Transformer speech enhancement models trained on 1 or 2 second clips keep most of their quality on 5 to 20 second inputs.
An exploration of length generalization in transformer-based speech enhancement,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
extension 1
citation-polarity summary
fields
eess.AS 1years
2025 1verdicts
CONDITIONAL 1roles
extension 1polarities
extend 1representative citing papers
citing papers explorer
-
Exploring Length Generalization For Transformer-based Speech Enhancement
A per-head learned linear bias on relative positions lets Transformer speech enhancement models trained on 1 or 2 second clips keep most of their quality on 5 to 20 second inputs.