Posts by Ian Goetz

HD Voice - Many issues to resolve

Before declaring the HD Voice is easy to do, you need to be sure that all aspects of the end to end experience have been covered. WB-AMR was standardised as a CODEC and NOT a service. Hence, all the end to end call path aspects were not fully considered at the standardisation stage, including the Handset Acoustic interface, which was under the remit of the same standards body, 3GPP SA4, as the codec. The solution relies on a clear end to end path and is only applicable for mobile to mobile calls. Hence the TFO and TrFO protocols are used. However, with echo cancellers, handovers and other in call processes to consider, the impact of changing from "HD Voice" to normal voice and back within a call has to be considered. In addition, the acoustic interface on the handset is more complex and has an impact on the capability of the device for normal voice. By widening the speech band you don't just pick up more voice, you pick up much more background noise as well. HD Voice will also suffer from a similar issue to 3G Video Calling - how do you know the far end party is able to process HD Voice, their is no voice capable presence enabkled phone book to tell you.

When you then add the long speech delays inherent in a mobile network to this clearer codec and all the end to end issues above, it is far from certain that HD Voice will be a success.

Call Quality on GSM and 3G

So, many of the comments noted that call quality has improved since 1991, EFR came out and was standardised in 1995-6 reaching networks in 1997. AMR provided the improved EFR quality over a much wider range of radio (C/I) conditions and could be used to offer that quality in a HR channel when radio conditions were good, thus saving significant amounts of GSM base station investment - which is why the main vendors of the time delayed it's incorporation into their products for as long as possible.

Other call quality improvements came from independenrt vendors using Signal Processing on the GSM A-Interface, notably Tellabs to reduce back ground noise transmitted from the mobile and remove acoustic echo from the mobile. Cellnet, now O2, in the UK implemented the system as Clear Call in 1999 on GSM. These problems were created by inadequate acoustic specifications in GSM standards which were carried through to 3GPP.

WB_AMR is built on top of the GSM & 3G AMR standards using the Transcoder Free Operation spec (a s/w upgrade to most TRAU equipment). The problem is that the acosutic specs were never upgraded to allow for WB-AMR and TFO drops from WB to standard NB for every handover. That combined with the increased psychoacoustic expectation that WB brings and the issues of transmission planning to reduced delay in an increasingly IP network and it is a very tricky technology to inrtroduce successfully.