Both the PESQ and POLQA algorithms for measuring audio quality are supported by the Cyara Platform, each has their own benefits and limitations. In this article we aim to detail the situations when using either algorithm will lead to the best results for your Cyara Test Cases & Campaign runs.
Both PESQ and POLQA generate a MOS (Mean Opinion Score), which is a 1-5 scale representing perceived audio quality for a human listener as below.
POLQA is the preferred algorithm for measuring voice quality on wideband networks, see below for a graph of common voice codecs and how they are defined as narrowband and wideband.
POLQA is the successor of the PESQ algorithm, but both have their own unique applications, see the table below for which algorithm is more appropriate for each scenario.
| Scenario | Preferred Method |
|---|---|
| Wideband Audio (7 kHz to 14 kHz frequency range) | POLQA |
| Narrowband Audio (300Hz to 3400 Hz voice bandwidth) | PESQ |
| "Adaptive" Audio Codecs (Codecs that have dynamic bitrates or noise reduction components) | POLQA |
| G.711 / VoIP Audio Codec Networks | PESQ |
| WebRTC Networks (see below) | POLQA |
As of the 21.4 Cyara Platform Release, the following Cyara products are enabled with POLQA for voice quality testing.
- ResolveAX
- Cyara Virtual Agent for the following WebRTC supported environments
- Genesys Cloud
- Amazon Connect
- NiceCXone (InContact)
- Twilio Flex
For more information on POLQA consult the POLQA Guidelines
For more information on PESQ, consult the PESQ Guidelines
Comments
0 comments
Please sign in to leave a comment.