“The 3-dB Transcoding Penalty in Digital Cellular and Interoperability with
Future Voice/Audio Coding Standards”
Dr. Jerry Gibson, UC Santa Barbara
Sponsored by the Dallas Chapter of the IEEE Signal Processing Society
In spite of the widespread attention to data and video, voice is still responsible for up to 75% of the revenue in wireless communications today. An unfortunate characteristic of 2nd and 3rd generation digital cellular systems has been the need to transcode at most network interfaces since the voice codec at the other end of the call is usually unknown and cannot be negotiated. Fourth generation systems such as LTE also require transcoding when the call leaves the LTE network. Transcoding at network interfaces adds complexity, degrades quality and increases latency, all of which directly affect the quality and cost of voice communications. We investigate the issues in voice communications over tandem connections of wireline and wireless communications links using rate distortion theoretic results and speech coding studies and show that each transcoding operation can incur a 3-dB penalty in source coding performance, in addition to increased latency and complexity. We also present several suggestions for addressing this performance. We then present characteristics of evolving standards for wideband (50 Hz to 7 kHz), superwideband (50 Hz to 14 kHz) and fullband (20 Hz to 20 kHz) speech/audio and how these codecs will interoperate with future wireless systems. This research was supported in part by NSF grants CCF-0728646 and CCF-0917230.
Jerry Gibson is chair of electrical and computer engineering at the University of California, Santa Barbara. His research interests include data, speech, image and video compression, multimedia over networks, wireless communications, information theory, and digital signal processing. He received the Fredrick Emmons Terman Award from the American Society for Engineering Education in 1990, and in 1992 he was elected an IEEE fellow “for contributions to the theory and practice of adaptive prediction and speech waveform coding.”