Codec artifacts are faint and need trained ears to detect

Evidence

There is a point of diminishing returns in the performance of all codecs; as the data rate is reduced each codec has a different data rate at which it begins to exhibit audible problems—artifacts. And, to make things more complex, different codecs exhibit different kinds of artifacts. It depends on the strategy used to identify components of the sound that are to be encoded more simply, or discarded. All codecs can be made to misbehave, but the best of them misbehave in a manner that is revealed as a momentary “difference” in sound, not a gross distortion or lapse in information. Most of the time, competent codecs are transparent. Those people involved with the subjective evaluation of codecs have found it necessary to train listeners to know what to listen for. It is not common for these problems to be discovered in casual listening. — , Toole, Floyd E - Sound reproduction-…, 1744

Categorization

• XREF: Parallel to the reader’s work on how training shapes perception (Deep Listening, trained listening). — connects to: Ear-level microphone measurements capture artifacts inaudible to human listeners — , Toole, Floyd E - Sound reproduction-…, 1355; Perceptual coders exploit masking to cut audio storage and bandwidth — David Howard, Jamie Angus, coustics and Psychoacoustics, 1656; Audio coding aims to reconstruct signals while other processing modifies them (+2 more)

Related

• Audio compression exploits masking to discard inaudible signal parts (similarity 0.82)
• Perceptual coding works by maximizing the surprise value of coded signals (similarity 0.80)
• Perceptual coders exploit masking to cut audio storage and bandwidth (similarity 0.80)
• Audio compression maximizes information by encoding only unpredictable signal parts (similarity 0.79)
• Lossless audio coding works by predicting and stripping signal correlations (similarity 0.79)
• Spectral masking by loud sounds is exploited to design lossy audio codecs. (similarity 0.78)
• Audio coding aims to reconstruct signals while other processing modifies them (similarity 0.78)

Broader Connections

• Ear-level microphone measurements capture artifacts inaudible to human listeners #corroborates (0.90) — , Toole, Floyd E - Sound reproduction-…, 1355
• Only trained, prompted ears can parse sound into individual partials #corroborates (0.80)
• Perceptual coders exploit masking to cut audio storage and bandwidth #expands (0.78) — David Howard, Jamie Angus, coustics and Psychoacoustics, 1656
• Audio coding aims to reconstruct signals while other processing modifies them #expands (0.72)
• Codec effectiveness trades off against bits spent on side information #expands (0.72)