Adding voicing and auditory features and adapting solo-singing models with limited polyphonic data reduces alignment errors for lyrics in polyphonic audio.
Performance on solo-singing In the first set of experiments, we explore the effect of each of the speech and music informed feature groups combined with MFCCs and i-vectors
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
eess.AS 1years
2019 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Acoustic Modeling for Automatic Lyrics-to-Audio Alignment
Adding voicing and auditory features and adapting solo-singing models with limited polyphonic data reduces alignment errors for lyrics in polyphonic audio.