StoRIR: Stochastic Room Impulse Response Generation for Audio Data Augmentation

• DOI: 10.21437/interspeech.2020-2261
• Publication Date: 2020-10-27T09:22:11Z
• Authors: Piotr Masztalski; Mateusz Matuszewski; Karol Piaskowski; Michal Romaniuk
• Keywords: Python; Impulse response

In this paper we introduce StoRIR -a stochastic room impulse response generation method dedicated to audio data augmentation in machine learning applications.This technique, contrary geometrical methods like image-source or ray tracing, does not require prior definition of geometry, absorption coefficients microphone and source placement is dependent solely on the acoustic parameters room.The intuitive, easy implement allows generate RIRs very complicated enclosures.We show that StoRIR, when used for a speech enhancement task, deep models achieve better results wide range metrics than using conventional method, effectively improving many them by more 5 %.We publish Python implementation online 1 .