VocalSet
A multi-singer vocal dataset containing professional recordings of various vocal techniques and singing styles, commonly used in research and evaluation for tasks such as singing voice analysis, technique recognition and emotion recognition.
Timeline (last 90 days)
The study conducts out-of-domain evaluation on the VocalSet dataset, examining the generalization ability of the features on broader singing data and revealing the connection between emotional expression and actual technique execution.
VocalSet is a public singing voice dataset recorded by professional singers under multiple sets of vowels, scales and technique conditions, covering various vocal techniques such as vibrato, pharyngeal voice, breathy voice and straight tone, and annotated with singer identity, technique type and pitch information.
The VocalSet dataset consists mainly of vocal-exercise-style singing, with limited coverage of repertoire and emotion labels, making it difficult to fully represent the complex emotional expression in real songs.
The study's conclusions come from a single research team's experimental setup, the out-of-domain evaluation is limited to only one dataset, VocalSet, and whether the interaction effect between vibrato and articulation rate holds stably in other singing styles remains to be independently verified.
All Facts (4)
The study conducts out-of-domain evaluation on the VocalSet dataset, examining the generalization ability of the features on broader singing data and revealing the connection between emotional expression and actual technique execution.
50%UnverifiedVocalSet is a public singing voice dataset recorded by professional singers under multiple sets of vowels, scales and technique conditions, covering various vocal techniques such as vibrato, pharyngeal voice, breathy voice and straight tone, and annotated with singer identity, technique type and pitch information.
50%UnverifiedThe VocalSet dataset consists mainly of vocal-exercise-style singing, with limited coverage of repertoire and emotion labels, making it difficult to fully represent the complex emotional expression in real songs.
50%UnverifiedThe study's conclusions come from a single research team's experimental setup, the out-of-domain evaluation is limited to only one dataset, VocalSet, and whether the interaction effect between vibrato and articulation rate holds stably in other singing styles remains to be independently verified.
50%