How do you take a sample of analog voice and digitize it? This is what a codec basically does, and there are several different codec flavors used in telecommunications networks to transport voice and video across IP networks. What is the reason for this multitude of coding/decoding (thus the term codec) approaches? The way that fixed and mobile operators have deployed their access and core IP networks can introduce constraints that impact their codec selection as can the type of service they are delivering.
Some codecs require less bandwidth than others making them ideal for connections that are constrained from a bandwidth perspective. However, these lower bit rate codecs can be computationally intensive and can have higher delay characteristics. In addition, an operator can adjust the packetization rate for a given codec to improve the packet handling performance of its network. Ultimately, fixed and mobile operators will select codecs that allow them to strike a balance between voice quality, bandwidth resources and network node performance. But for end-to-end service continuity within and between networks, operators must somehow make these different encoding and packetization methods interwork. This is where transcoding and transrating come into the picture. Transcoding is used to convert one codec type to another and transrating is used to covert one packetization rate to another.
At one time the need for transcoding was relatively basic having to only accommodate a small set of codecs, however, the number of codecs in use has exploded. Codecs specified for use by the Third Generation Partnership Project (3GPP) not only reduced the amount of bandwidth needed compared to the earliest codecs like G.711, but also supported multiple allowable bit-rates within the same codec. Adaptive Multi-Rate Narrow Band (AMR-NB) and Adaptive Multi-Rate Wide Band (AMR-WB) are now the standard codecs for many mobile devices and voice networks. AMR-WB delivers High Definition quality voice (HD Voice) which enhances the listener experience over that provided by AMR-NB. Another HD quality voice codec called Enhanced Voice Services (EVS) has been introduced that interoperates well with AMR-WB on all rates.
You would think that as networks mature, the number of different codecs used by end points would converge, but this is not the case. Codec types used in modern fixed and mobile networks and Over the Top (OTT) networks continue to expand and may use various packetization rates to optimize network node performance. Voice over LTE (VoLTE) networks which use an IP Multimedia Subsystem (IMS) core to enable voice services will continue to drive the need for transcoding between IP networks but also with PSTN handoffs. While a network may have to support a long list of codec types, the different codecs are to an extent network technology dependent; for example:
- VoLTE/IMS: EVS, AMR-WB, AMR-NB, H.263 and H.264
- 3G/UMTS: AMR-NB and G.711
- 2G/GSM: GSM EFR and G.711
- OTT: iLBC (Google talk, Yahoo Messenger), Opus (WebRTC), SILK (Skype), H.263 and H.264
So it’s more important than ever that a service provider develop a comprehensive transcoding and transrating strategy that takes into account:
- Flexibility to offer new services by supporting either legacy or modern voice and video codecs
- Intelligent network architecture to centralize, distribute or virtualize transcoding functionality appropriately
- Robust quality of service that minimizes the number of transcoding events while balancing the bandwidth needed
We recently put together a white paper that you can download to help you overcome challenges with end-to-end service continuity called, “Which Media Transcoding Strategy Is Right for Your Network?” Check it out and let us know what your transcoding challenge was and how you resolved it by tweeting us @Dialogic.