Vincent Lostanlen
@lostanlen.bsky.social
Scientist at CNRS.
https://audio.ls2n.fr
A science game to test your musical memory: https://tunetwins.app
https://audio.ls2n.fr
A science game to test your musical memory: https://tunetwins.app
(part of) the Audio@LS2N research group, one year later
Léo Coquet
Hugo Ballesteros
Matthieu Carreau
Yasmine Benhamadi
VL
Orian Sharoni
Ainė Drelingytė
Xiran Zhang
audio.ls2n.fr
Léo Coquet
Hugo Ballesteros
Matthieu Carreau
Yasmine Benhamadi
VL
Orian Sharoni
Ainė Drelingytė
Xiran Zhang
audio.ls2n.fr
November 14, 2025 at 6:44 AM
(part of) the Audio@LS2N research group, one year later
Léo Coquet
Hugo Ballesteros
Matthieu Carreau
Yasmine Benhamadi
VL
Orian Sharoni
Ainė Drelingytė
Xiran Zhang
audio.ls2n.fr
Léo Coquet
Hugo Ballesteros
Matthieu Carreau
Yasmine Benhamadi
VL
Orian Sharoni
Ainė Drelingytė
Xiran Zhang
audio.ls2n.fr
granta.com/common-cyborg/ "Common Cyborg" by Jillian Weise
in "Disability Visibility" (dir. Alice Wong)
in "Disability Visibility" (dir. Alice Wong)
October 15, 2025 at 7:40 PM
granta.com/common-cyborg/ "Common Cyborg" by Jillian Weise
in "Disability Visibility" (dir. Alice Wong)
in "Disability Visibility" (dir. Alice Wong)
un cadeau de l'auteur
October 12, 2025 at 6:05 PM
un cadeau de l'auteur
We also prove that STONE induces spontaneous symmetry breaking, since modes of limited transposition (e.g., augmented, diminished, whole-tone) are multistable in the circle of fifths, the associated CPSD gradient is driven towards more "tonal" (i.e., asymmetric) scales, e.g., diatonic or pentatonic.
October 9, 2025 at 7:19 AM
We also prove that STONE induces spontaneous symmetry breaking, since modes of limited transposition (e.g., augmented, diminished, whole-tone) are multistable in the circle of fifths, the associated CPSD gradient is driven towards more "tonal" (i.e., asymmetric) scales, e.g., diatonic or pentatonic.
Our main result. If the frequency of cross-power spectral density is set to be coprime with Q=12, STONE does not collapse: the self-supervised objective is well-posed and sparsity-inducing in pitch class space.
October 9, 2025 at 7:13 AM
Our main result. If the frequency of cross-power spectral density is set to be coprime with Q=12, STONE does not collapse: the self-supervised objective is well-posed and sparsity-inducing in pitch class space.
"Understanding equivariant self-supervised learning in musical pitch class space" is out on IEEE Signal Processing Letters!
in which we revisit STONE (Kong et al., ISMIR 2024) from a more theoretical standpoint
🔗 hal.science/hal-05297975v1
in which we revisit STONE (Kong et al., ISMIR 2024) from a more theoretical standpoint
🔗 hal.science/hal-05297975v1
October 9, 2025 at 7:07 AM
"Understanding equivariant self-supervised learning in musical pitch class space" is out on IEEE Signal Processing Letters!
in which we revisit STONE (Kong et al., ISMIR 2024) from a more theoretical standpoint
🔗 hal.science/hal-05297975v1
in which we revisit STONE (Kong et al., ISMIR 2024) from a more theoretical standpoint
🔗 hal.science/hal-05297975v1
