eduzhai > Applied Sciences > Engineering >

AMRConvNet AMR-Coded Speech Enhancement Using Convolutional Neural Networks

  • king
  • (0) Download
  • 20210507
  • Save

... pages left unread,continue reading

Document pages: 6 pages

Abstract: Speech is converted to digital signals using speech coding for efficienttransmission. However, this often lowers the quality and bandwidth of speech.This paper explores the application of convolutional neural networks forArtificial Bandwidth Expansion (ABE) and speech enhancement on coded speech,particularly Adaptive Multi-Rate (AMR) used in 2G cellular phone calls. In thispaper, we introduce AMRConvNet: a convolutional neural network that performsABE and speech enhancement on speech encoded with AMR. The model operatesdirectly on the time-domain for both input and output speech but optimizesusing combined time-domain reconstruction loss and frequency-domain perceptualloss. AMRConvNet resulted in an average improvement of 0.425 Mean Opinion Score- Listening Quality Objective (MOS-LQO) points for AMR bitrate of 4.75k, and0.073 MOS-LQO points for AMR bitrate of 12.2k. AMRConvNet also showedrobustness in AMR bitrate inputs. Finally, an ablation test showed that ourcombined time-domain and frequency-domain loss leads to slightly higher MOS-LQOand faster training convergence than using either loss alone.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×