CITATION — REFERENCE ENTRY

spectrogram-masking-ahead · defossez2019-demucs

Revision c459c7be-92b8-4dbe-9871-b52677b9d46b · 11/21/2026, 6:53:18 AM UTC
Claim ID
spectrogram-masking-ahead
Assertion
At SiSEC 2018, the best waveform-predicting model averaged an SDR of 3.2 over four sources, against 5.3 for the best spectrogram-masking approach.
Quote
in the last SiSec Mus evaluation campaign [Stöter et al., 2018], the best model that directly predicts waveforms achieves an average signal-to-noise ratio (SDR) over all four sources of 3.2, against 5.3 for the best approach that predicts spectrograms masks
Quote language
en
Locator
page: 2 (sec. 1)
Available in