Hello, I have been playing with the castanet sample for a long time now as I know how bad all DCT codecs performs on this specific sample, & I was always surprised how quality was falling drastically between nero aac 96Kbps LC & nero aac 64Kbps HE ... as I was not using nero aac as my main codec I didn't care much at the time ... I recall I even read several enthousiasts on HA telling how increadibly good HE aac 48Kbps was for the bitrate ... so far I shut up & told myself that those people must be deaf ... but I always thougth something was wrong with SBR on music ... I have no problem with using AAC HE on voice, but for me using AAC HE for music was a NO-NO due to my ABX experience with the castanet sample ==> castanet at q0.25 (with SBR by default) is HORRIBLE. No discussion possible, if you don't hear the massacre, you're deaf.

As foobar2000 nero aac interface doesn't allow the force LC setting so far I was always too lazy to test castanet at 64Kbps LC ... but today I did it ... & BOOM what I was suspecting is true ... nero 64Kbps with force LC is much much better than nero 64Kbps with SBR which is the default nero setting ... at least on the castanet sample.

So I decided to upload the samples so that anyone can compare ... I can ABX it 100% in less than 2 min which means instantly ...

So I am asking if anyone can provide me with a music sample where nero aac 64Kbps with SBR is better than nero aac 64Kbps with force LC, to justify the use of SBR at 64Kbps

If nobody can provide me with such a sample, I am asking nero developers to remap the use of SBR to settings below 64Kbps because it's simply worst at this bitrate for music ... & if I wanted to use nero aac HE for voice I would use 48kbps anyway.

Thks for listening. Hope it helps.

Edit: Plz listen to original first then compare to q0.25 with SBR to hear the massacre, then listen to q0.25 lc (without SBR) to see how it is increadibly better than the aac he q0.25 default setting...

You want us to make decisions about something we've been (fine)tuning for years based on one sample? A sample that is commonly known to all developers working on audio codecs. A castanets sample does not represent the whole of "music"
Try any other sample yourself, you will notice that in by far most cases HE AAC is better than LC AAC at 64 kbps.

You've made one known mistake, and one unknown (to you) mistake:


First:
Castanets_64Kbps_q0.25_sbr.mp4 : 67 kbps
Castanets_64Kbps_q0.25_lc.mp4 : 77kbps

that's more than 10% of difference in bitrate.

And the explanation (and the second mistake) was given several times by Nero developers:

when you add the "-lc" or "-he" or "-hev2" parameters, you're changing the scale, and no longer corresponds a -q x.x to another -q x.x.

The reason of this is because it would not have sense (nor be possible) an -hev2 that gave the same bitrate at -q 10 than an -lc file, nor it is expected to give the same quality (because you're forcing a mode of operation).


There are people that dont like SBR, that even hate it. It is an abxable effect, which usually is acceptable over an overall reduction of quality ( a 64kbps CBR LC file, for starters, has a lower sampling rate. Of course there are more reasons than this one)

well I don't use Nero AAC so I won't argue with the developpers if they are deaf

but I want to clarify things:

- I know I changed the scale.
- I know the bitrate is higher.
- I know one sample is just one sample.
- I tested Nero AAC on normal music at near 64Kbps I couldn't make my opinion based on ABX faulty.

even knowing all this, & knowing you both are respectable members, I believe I am 100% right, & as I am sure I am right I couldn't resist posting. I just though the difference in quality between the two samples was SO OBVIOUS that it would be instantly clear to you that there is a problem (even with 10% bitrate difference). Sorry if my naiveness made you waste your time.

To be honest, I was expecting such an answer because I know you from a long time reading your posts.
I am pretty sure for example that if my nick was Guruboolez you wouldn't have answered me the same way.

That's exactly why I use lossless instead of nero aac despite the fact that I think nero aac is the state of the art lossy codec. I don't want to be locked by irrationnal developper choices.

Discussion closed. Switching back to lossless mode. All apologies

Nobody claimed in this thread that HE is better than LC on this sample at 64kbps, and especially nobody claimed that you (sauvage78) don't hear the difference.
Decision which profile is going to be used at -q0.25 is based on the expected score of HE and LC on representative ~20 samples judged by 20-40 listeners (not deaf listeners, but listeners that can hear difference at this bitrate).
It can be expected that there will be a listener who will prefer LC on all of those samples. Definitely there will be cases where some user on ONE of those samples will think (correctly) that for him LC is way better than HE.
But in AVERAGE on all samples among all these listeners it can definitely be expected that HE will have better score.
Since you are not AVERAGE, but ONE, my advice is if you want to use 64kbps to test music that you usually listen and decide whether to use default HE or force LC.