Audio samples for "iMetricGAN: Intelligibility Enhancement for Speech-in-Noise using Generative Adversarial Networks based Metric Learning"


Authors: Haoyu Li, Szu-Wei Fu, Yu Tsao, Junichi Yamagishi

Submitted to INTERSPEECH 2020


Language: German (SNR = -9 dB)

Algorithms Speech (at same RMS) Speech-in-Noise
Plain
OptSII
OptMI
SSDRC
SiibGAN-zs
SiibGAN
MultiGAN

Language: English (SNR = -7.5 dB)

Algorithms Speech (at same RMS) Speech-in-Noise
Plain
OptSII
OptMI
SSDRC
SiibGAN-zs
SiibGAN
MultiGAN

Language: Spanish (SNR = -12.5 dB)

Algorithms Speech (at same RMS) Speech-in-Noise
Plain
OptSII
OptMI
SSDRC
SiibGAN-zs
SiibGAN
MultiGAN


Acknowledgement:
Speech materials were provided by The Hurrican Challenge