How do Lame mp3 and aac compare these days!

[Anonymous5394]

Both great audio formats and in widespread use, but how do they compare to each other in 2020? Is 128kbps aac equivalent to Lame 192kbps vbr? We all have varying degrees of hearing ability, but the notion of audio transparency is still a personal thing and the hardware you use at home. Who has magical hearing which can pick up those higher audio frequencies?

192kbps Lame sounds pretty good played via a Prologic II based TV, as does aac, and 192kbps aac is transparent enough for me for movies etc, but he-aac sounds tinny below bitrates of 64kbps and a no go for music imho, though some may differ. What is transparent for you?

[abolibibelot]

Someone(s) may chime in over here, but the best place to get extra precise and thorough replies on that kind of questions would have to be https://hydrogenaud.io/, where these kinds of debates are ignited on a nearly daily basis and discussed almost to death.

For my part, what I consider silly is the current widespread use of MP3 at 320kbps CBR, which is both overkill from a lossy compression perspective, and still degraded compared with lossless compression, which in some cases yields comparable or even inferior average bitrates... (Solo piano works for instance often compress to less than 300kbps in FLAC, which is not even the most efficient lossless codec.)

[ItaloFan]

No idea what the state of the art is in AAC encoding these days.

For MP3, LAME has had no quality related changes to the code in quite a while (though it seems some may be coming in 3.100.1?). Quality is measured in several ways, and making a change to improve one thing can result in a tradeoff in others. There is also no real framework for regression testing—somewhat understandable, given the subjective and fatiguing nature of listening tests.

I also think there's a bit of hubris and burnout among the developers. The lame-dev list is a ghost town, with the developers' interest only being mildly piqued by crash reports and a long-term attempt to disentangle the little-used decoder from the source code. By the mid-2000s, they got ~100 kbps and up to handle problem samples as best they could; and they're naturally unconcerned about occasional second-place showings in listening tests. All the major encoders are roughly equal in performance, at bitrates near transparency. Bitrates below 100 have received relatively little attention, despite the fact that a major quality boost could be obtained if they'd just replace the resampling code with something relatively noise-free.

Transparency typically happens for most people at bitrates far lower than the bitrates we believe are necessary, and nobody's hearing gets better over time, so our transparency threshold inevitably has dropped in the last 20 years. Nevertheless, I admit I am just as guilty of erring quite a bit on the high side—"better safe than sorry!"

ABX testing is the only way to be sure where your threshold is. foobar2000's built-in ABX comparator is great for this purpose, though you do still have to prepare good samples in advance. You basically have to correctly identify 11 out of 12, or it's statistically insignificant, since 6 out of 12 is no better than a coin toss. This also controls for inconsistent perception, e.g. where you hear differences that don't objectively exist.

I would not worry about what other people claim is transparent for them. Use what works for you. It seems you're already doing that. No need to second-guess everything!

[Anonymous5394]

Well said from both of you. The site mentioned hydrogenaud is a goldmine of information, but the debate gets very heated at times. I wanted to pose the questions here as it is an even more tempered site!

Transparency is a subjective thing and many cannot distinquish Lame 128kbps from mp3 320kbps, but the playback medium is a big factor too. I'd done some abx tests and found 168kbps to 170 aac to be the threshold of perceived transparency and is sharper though noisier than mp3. Lame comes in for me just a little higher on average for quality playback etc, and 128kbps has never ever seemed dynamic enough for me, especially in music, which I felt always deserved the minimum of 256kbps vbr or indeed losslessness in more difficult pieces e.g Orchestral. Nowadays my perceptual quality might be lower as hearing certainly detiorates with age! Thanks guys.

[abolibibelot]

Nowadays my perceptual quality might be lower as hearing certainly detiorates with age!

Depends, it's also subjective in a way... Even if the physical perception deteriorates, the qualitative appreciation of fine nuances in music can improve over time, while the desire to discover new sounds dwindles more often than not — the net effect of those antagonist tendencies being quite unpredictable. Reminds me of a brief text from famous french comedian Pierre Desproges (1939-1988), quite atypical for him as it's tender and nostalgic while he was known and renowned for his acerbic and misanthropic wit (somewhat akin to the late George Carlin).
“When I was still almost little, in the country, I would wait until the summer night became very dark, to put the speaker of my Teppaz atop the big lime tree, then I would listen to the Notte while counting the stars, lying in the grass, and waves of voluptuous sorrow would run across my skin, like when you're seperated from the one you love and it's so painful for ordinary joys. Today, I have a big 2 x 100W hi-fi system. When we grow up, our ears are sharpening and our hearts are shrinking.”

{Original french : “Quand j'étais presque encore petit, à la campagne, j'attendais que la nuit d'été fût très noire pour installer le haut-parleur de mon Teppaz en haut du grand tilleul, et j'écoutais la Notte en comptant les étoiles, couché dans l'herbe, et des vagues de chagrin voluptueux me couraient sur la peau, comme quand on est loin de l'autre qu'on aime et que c'est déchirant pour les joies ordinaires. Aujourd'hui, j'ai une maxi-chaîne deux fois cent watts. En grandissant, l'oreille s'affine et le cœur se serre.”}

[Anonymous5394]

Immediately I could see both an irony and comparison in that expresive French quote to music itself and to allusions of translation. Music compressed to any format loses its purity though not its intention, but purists will still prefer the original(wave or indeed vinyl) to appreciate more the essence of that purity and its heart.

Translation from an original piece into English as you give does confer the meaning very well, but the French captures the essence and understanding on a deeper level that much better. Sublime! This is true of anything in any original language format be it music or human speech.