Different CODECs produce different outputs of varying quality traded against bandwidth utilization. The choice of which CODEC to use is based on a number of of things including computational load placed on the devices processing the audio, the desired quality of the audio, the network bandwidth required to transport it, and interoperability with systems that are limited to using certain CODECs.

CODEC design must provide a balance between these conflicting goals: How to represent audio in its highest fidelity, using a few bits as possible, with as little burden on the end-user device CPUs as possible ('computational complexity'). Historically, computational complexity was always a big factor as CPUs were pretty slow by today's high-performance hardware standards, and, therefore, engineers often chose CODECs that would not tax hardware. Thus, we have a historical proliferation of CODECs (such as G.711) that uses 64 Kpbs of bandwidth with very low computational complexity. With newer hardware, we're free to choose more computationally intensive CODECs that provide a better audio experience with reduced bandwidth requirements.

To wit: G.711 is fixed at 64 Kbps and produces high-quality audio at low complexity. Opus produces audio quality that is even better than G.711 with a much-reduced bandwidth footprint. In fact, Opus operating at 16kbps produces higher audio quality than G.711 requiring four times the bandwidth. With ICE, you can go all the way down to 6kbps on Opus.

At first glance, you might assume that someone transmitting audio using an Opus 16 CODEC is going to use 16kbps on the network. But there's a whole lot of devil in the details.