Fig. 4From: Stripe-Transformer: deep stripe feature learning for music source separationThe average score of SDR, SIR, and SAR on “vocals,” “drums,” and “bass” categories with different input audio lengthsBack to article page