MOS, PESQ, and POLQA are all methods used to measure and define audio quality in telecoms. MOS focuses on the metrics associated with internal network traffic, while PESQ and POLQA compare and analyze the audio captured on the complete end-to-end call path.
The Differences Between MOS, PESQ, and POLQA
| Parameters | MOS | PESQ | POLQA |
|---|---|---|---|
| Subjective Results (Based on network performance) |  |
|
|
| Support for Standard-voice (Narrowband 300 – 3,400 Hz) | |
|
|
| Objective Results (Based on the real audio analysis) |  |
|
|
| Analyzes the Entire Call Path (Considers all aspects of the call path from start-to-end) | |
|
|
| Repeatable Results (Results are repeatable with objective analysis) | |
|
|
| Support for HD-voice (Wideband 50 – 7,000 Hz) | |
|
|
| Support for Codec Comparison (Compare narrowband and wideband results accurately) | |
|
|
| Support for Codec Comparison (Compare narrowband and wideband results accurately) | |
|
|
What are the benefits of PESQ and POLQA?
- Analysis of the entire call path: PESQ and POLQA analyze the whole call path rather than just the last hop, which allows these algorithms to reflect the overall quality of the audio stream more accurately. In contrast, MOS only considers the last call leg and may not accurately reflect the overall quality of the audio stream.
- Objectivity: PESQ and POLQA are objective algorithms that analyze the actual audio signal, rather than relying on assumptions of quality which are applied based on the metrics and performance of the internal network. This means that their results are more accurate and repeatable.
- HD-voice Support: Within wideband and super wideband frequency ranges PESQ and POLQA are able to evaluate the performance of codecs up to 14,000Hz, while MOS is limited to narrowband speech and is not suitable for evaluating newer codecs.
Where does MOS lack in modern telecoms testing?
- When you use MOS you look at the network metrics and determine the audio quality from it without actually trying to “listen” to the audio quality itself.
- With MOS you rely only on the last call leg where MOS was calculated, missing any call quality degradation that occurred further downstream before it hit the network.
- When you use MOS you focus on narrowband speech only, missing out on HD voice communication characteristics.
Why are these things important?
- Catching issues in the entire call path: Considering the whole call path from the speaker’s side to the listener’s side will provide a much more accurate representation of the call quality as it can help identify quality issues that occur on the public switched telephone network (PSTN) before the audio reaches your internal network.
- Staying objective: Objective testing methods such as PESQ and POLQA consider only the facts related to the actual audio signal captured on the test call. This kind of audio quality analysis means that if your audio quality score changes you will know that there has been a change in the call setup or that transcoding has taken place somewhere on the call path.
- Considering HD-voice: To accurately predict the audio quality of voice signals on the wideband frequency range, it is necessary to use the current audio quality testing standards. The PESQ and POLQA algorithms are optimized for use with modern audio codecs and can accurately compare the performance of different codecs and frequency ranges. These algorithms are able to accurately predict the audio quality of voice signals in the wideband frequency range by taking into account the characteristics of the codecs being used.
Why are PESQ/POLQA better?
PESQ and POLQA can provide accurate measurements of the full call path from the external to the internal network. If you want to deliver an excellent customer experience from end to end with effective representations of the newer audio codecs, it is highly advisable to use PESQ or POLQA over MOS.
MOS has the advantage of being quick to set up without using a reference audio signal to compare audio against. Still, the drawbacks of this method are that it produces less accuracy and results are not repeatable. A “non-intrusive” approach that relies solely on single-sided monitoring without creating traffic may occasionally be better from the viewpoint of a network operator if the focus is on network quality control.
Summary – Key Points
- PESQ and POLQA are generally considered more accurate and repeatable methods for evaluating the quality of audio streams than MOS.
- PESQ and POLQA are objective algorithms that analyze the actual audio signal rather than relying on assumptions of quality which are applied based on network performance.
- PESQ and POLQA analyze the entire call path, rather than just the last hop, and can produce both subjective and objective results. In contrast, MOS is a subjective measure of audio quality based on the last call leg, which may not accurately reflect the overall quality of the audio stream.
- MOS only considers narrowband speech and is not well-suited for evaluating the performance of newer codecs or emerging voice technologies.
For more information read The Evolution of MOS.