MOS-LQO: definition and a practical example
Short answer: MOS-LQO is the 1-to-5 score, defined in ITU-T P.800.1, that an algorithm calculates by comparing the audio received on a call with the reference audio that should have arrived.
In Portuguese: MOS-LQO
What is MOS-LQO?
MOS-LQO (Mean Opinion Score, Listening Quality Objective) is a 1-to-5 score that an algorithm calculates from the actual audio of a call. The algorithm takes two files, the reference audio and the audio that arrived at the other end, and measures how far the second has drifted from the first. The result is expressed on the same scale as the ITU-T P.800 MOS, where 5 is excellent and 1 is bad.
The acronym comes from ITU-T P.800.1, which classifies MOS scores by their source. “LQ” stands for listening quality and “O” for objective, meaning calculated by a machine from the sound. It contrasts with LQS, given by listeners, and with estimated scores, such as the E-model’s MOS-CQE, which never listen to the audio.
Measured or estimated
The practical difference is what each score can see. An estimated score looks at the packets: delay, loss and jitter. A measured score looks at the sound. If the call arrives silent, if only one side has audio, if the carrier plays a message instead of reaching the destination or if a transcoder distorts the voice, the packets can be perfect and the estimated score stays high. MOS-LQO drops, because the received sound does not resemble what was expected.
The best-known algorithms for MOS-LQO are PESQ and POLQA, both ITU standards, and the open-source ViSQOL.
In practice
Picture three calls to the same number, all with 0% packet loss and jitter under 5 ms:
| Call | What happened | Network | MOS-LQO |
|---|---|---|---|
| A | Clean audio | Normal | High |
| B | Audio went through two transcoders | Normal | Lower |
| C | Carrier played “number unavailable” | Normal | Very low |
On the network, the three calls are identical. In the audio, only call A delivered what it should. That difference is why it pays to measure instead of estimate.
How Monitor uses this
Monitor measures MOS-LQO on every answered test call. It compares the received audio with the expected recording, the reference recorded when the circuit is set up, using ViSQOL in speech mode. The default thresholds: from 3.8 up the call is healthy; below 3.0 it counts as degraded; between 3.0 and 3.8 it passes with a note. See voice quality with MOS-LQO.
Frequently asked questions
What is the difference between MOS-LQO and MOS-CQE?
MOS-LQO is measured: an algorithm compares the received audio with the original audio and produces the score. MOS-CQE is estimated: the ITU-T G.107 E-model calculates the score from delay, loss and codec, without listening to anything. That is why MOS-CQE can give a good score to a call that landed on a carrier recording, and MOS-LQO cannot.
Is a MOS-LQO from one algorithm valid for another?
Not directly. PESQ, POLQA and ViSQOL produce scores on the same 1-to-5 scale, but each has its own mapping and its own ceiling. ViSQOL in speech mode reaches 5.0, while audio mode tops out near 4.75. Compare scores from the same algorithm, in the same mode, before drawing conclusions.