Dimitrios Bralios @DBralios

Great audio AEs/codecs exist, but when you need structured latents or a tweaked bottleneck for a downstream task (e.g. generation), retraining is expensive & brittle. We Re-Bottleneck👇

arXiv Sound @ArxivSound

10 months ago

Dimitrios Bralios, Jonah Casebeer, Paris Smaragdis, "Re-Bottleneck: Latent Re-Structuring for Neural Audio Autoencoders," arxiv.org/abs/2507.07867

0 1 5 3K 7

1 5 21 3K 13

View Details

Audio and Speech Processing Papers @AudioAndSpeech

12 months ago

Re-Bottleneck: Latent Re-Structuring for Neural Audio Autoencoders. arxiv.org/abs/2507.07867

0 2 6 383 1

View Details

Dimitrios Bralios @DBralios

a year ago

@gmongaras This recent paper explores this question arxiv.org/abs/2502.03444

1 0 1 81 1

View Details

Dimitrios Bralios @DBralios

3 years ago

You can find the pre-print here: arxiv.org/abs/2310.10604

0 1 0 829 0

View Details

I'm thrilled to announce that our paper, "Generation or Replication: Auscultating Audio Latent Diffusion Models" 🩺 with the Speech & Audio team at MERL, has been accepted for publication at #ICASSP2024!

1 1 14 1K 1

View Details

arXiv Sound @ArxivSound

3 years ago

``Generation or Replication: Auscultating Audio Latent Diffusion Models. (arXiv:2310.10604v1 [eess.AS]),'' Dimitrios Bralios, Gordon Wichern, François G. Germain, Zexu Pan, Sameer Khurana, Chiori Hori, Jonathan Le Roux, ift.tt/FTuqsxK

0 4 9 2K 5

View Details

DailyAudioPapers @mlsp4audio

3 years ago

Complete and separate: Conditional separation with missing target source attribute completion Dimitrios Bralios, Efthymios Tzinis, Paris Smaragdis arxiv.org/abs/2307.14609

0 1 3 800 3

View Details

Efthymios Tzinis @ETzinis

6 years ago

A little bit of bias by my side 🎶! We often neglect the variety in importance of training examples for separation models. Interestingly, we can solve important problems like robustness! Please check our work with @DBralios! arxiv.org/pdf/2010.13228…