Skip to content

CONCEPT Cited by 1 source

POLQA MOS metric

Definition

POLQA (Perceptual Objective Listening Quality Analysis, ITU-T P.863) is an objective voice-quality metric that outputs a MOS (Mean Opinion Score) on a 1–5 scale by comparing a decoded audio signal against a reference, modelling human perceptual sensitivities. Higher is better. POLQA is the standard ITU benchmark used to compare codecs (Source: sources/2024-06-13-meta-mlow-metas-low-bitrate-audio-codec).

Scale

Standard MOS interpretation (ITU-T P.800 family):

MOS Quality label
5 Excellent
4 Good
3 Fair
2 Poor
1 Bad

Used in the MLow post

The central comparison:

Codec / bitrate POLQA MOS
Opus @ 6 kbps NarrowBand 1.89
MLow @ 6 kbps WideBand 3.9

Meta characterises this as "two-times better quality than Opus". Note: this is the MOS ratio 3.9 / 1.89 ≈ 2.06not a doubling in perceived-quality units. MOS is non-linear in perceived quality; a 1.89 → 3.9 shift is roughly "barely intelligible" → "good" on the subjective label scale.

Why POLQA (not PESQ)

POLQA (P.863, 2011) supersedes PESQ (P.862, 2001) and explicitly targets HD / WideBand / SuperWideBand audio, which PESQ handled poorly. For codecs that move between band modes (see concepts/narrowband-vs-wideband-audio), POLQA is the correct benchmark.

Caveats named by the source

The MLow post does NOT disclose:

  • The reference signal set used.
  • Whether the scores are mean or median across a corpus.
  • Test conditions (single-channel, multi-channel, background noise, etc.).
  • The full POLQA vs bitrate curve across operating points (Figure 2 is shown in the post but without tabulated numbers).

Seen in

Last updated · 319 distilled / 1,201 read