Description

V3. Model trained using Papa Emeritus’s isolated vocal tracks from the releases Opus Eponymous, Secular Haze (Vinyl Single), Infestissumam, Meliora, Popestar — EP, Ceremony And Devotion, and Prequelle by Ghost. All stems have had noise, echo, and reverb removed and were compressed + normalized to -1 dB for the sake of achieving an even & consistent sound. This proved to be a difficult & incredibly frustrating model to make… due to the mass amounts of reverb & echo that Papa Emeritus uses on every Ghost release, my typical methods of extracting vocals struggled to provide dry stems. It took a long time to figure out workarounds for the dataset and I’m still not completely satisfied with the results. Hopefully it sounds okay!