This repository provides the dataset we used in the paper “IMPROVING LOW-COMPLEXITY AND REAL-TIME DEEPFILTERNET2 FOR PERSONALIZED SPEECH ENHANCEMENT”, termed simulation-set, which can be found in https://drive.google.com/drive/folders/1yrye3yejh6Hnmn_g4JPxXXAkFsbyY3a8?usp=sharing.
The three folders contain noisy speech, clean speech and enrollment speech respectively, and the corresponding relationship between them are recorded in the SCP files. In addition, we provide several noisy and enhanced samples in pDF2 for listening.