Customer ExperienceMonitoringSIPTestingUnified CommunicationsVoIPVoLTEWebRTC

What is PESQ (Perceptual Evaluation of Speech Quality)?

Speech Quality

PESQ is a testing and monitoring related acronym that stands for Perceptual Evaluation of Speech Quality. Typically pronounced “pesk,” it is used as a means for automating the assessment of speech quality in communications networks (Unified Communications, VoIP, etc.).

Previously, I outlined Mean Opinion Score (MOS) and described why it has become the de facto standard measure of speech quality in Enterprise communications environments. Although MOS is the most accurate measure of how people perceive speech quality, it has one major drawback: you need to actually ask people what they think. That makes it impossible to automate the calculation of MOS values.

Read: Top Ten Tips for Testing Your Contact Center Network

That’s exactly why PESQ was developed and standardized.

The PESQ Standard

PESQ, defined by ITU-T recommendation P.862, allows you to predict how people would perceive voice quality if they were listening in to a call. PESQ defines an intrusive method of measurement. I use the word “intrusive” because you need to start with a known voice clip, play it through the device or network in question, and capture the voice clip at the other end.

In practice, this process first involves establishing a session between two cooperating endpoints using a suitable protocol (SIP, H.323, WebRTC, etc.). Once the session is established, the endpoints execute scripted call flows to send known voice clips through the system and record them at the other end. After you have the original voice clip and a corresponding voice clip that has passed through the system (device, network, etc.), you’re ready to measure voice quality.

Read: Are Customers Seeing Problems Before You Do?

The heart of PESQ is a sophisticated algorithm that predicts the MOS value that would result if real people were evaluating the recorded speech clips. Simple metrics, such as signal-to-noise-ratio, have proven ineffective at predicting user experience. PESQ solves this by incorporating a perceptual model that can distinguish between audible and inaudible distortions that can occur in communications networks.

The benefits are twofold:

  1. It allows you to automate voice quality measurement in such a way that you can be confident that the scores accurately reflect what people would perceive.
  2. It’s repeatable. A change in PESQ scores means that something has changed on the network. In contrast, having no supporting data other than someone telling you “I think this sounded better before the upgrade” doesn’t give you a lot to work with.

Testing and Monitoring Communications Systems

PESQ is applicable to both testing and monitoring communications networks. Testing is done in lab environments where the system under test can range from a single device to an entire network. Either way, the tester can be assured that repeated tests will give consistent results.

Read: Take a Closer Look: What is Passive Monitoring?

PESQ is also very effective for monitoring speech quality in live networks. Note that you must use active monitoring techniques due to the intrusive nature of PESQ.

Are you monitoring your communications systems to ensure the best customer experience possible? Let me know about your experiences in the comments below.