eduzhai > Applied Sciences > Engineering >

A Deep Learning Approach for Low-Latency Packet Loss Concealment of Audio Signals in Networked Music Performance Applications

  • king
  • (0) Download
  • 20210506
  • Save

... pages left unread,continue reading

Document pages: 8 pages

Abstract: Networked Music Performance (NMP) is envisioned as a potential game changeramong Internet applications: it aims at revolutionizing the traditional conceptof musical interaction by enabling remote musicians to interact and performtogether through a telecommunication network. Ensuring realistic conditions formusic performance, however, constitutes a significant engineering challenge dueto extremely strict requirements in terms of audio quality and, mostimportantly, network delay. To minimize the end-to-end delay experienced by themusicians, typical implementations of NMP applications use un-compressed,bidirectional audio streams and leverage UDP as transport protocol. Beingconnection less and unreliable,audio packets transmitted via UDP which becomelost in transit are not re-transmitted and thus cause glitches in the receiveraudio playout. This article describes a technique for predicting lost packetcontent in real-time using a deep learning approach. The ability of concealingerrors in real time can help mitigate audio impairments caused by packetlosses, thus improving the quality of audio playout in real-world scenarios.

Please select stars to rate!


0 comments Sign in to leave a comment.

    Data loading, please wait...