• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

AAC's quality...

GrO

Member
Joined
Nov 23, 2021
Messages
49
Likes
5
Location
Poland
So, let's go

First :
- did I say that AAC 320 is in the developer statements, or would be better than AAC 256 ?
Never
, I only told you that Tidal was using AAC 320 VBR until now
(which may change as they start to use 160 now that they are seperating turning Hifi into AAC+FLAC(and MQA 16bit) and Hifi Plus AAC+FLAC+MQA (16 and 24bit)+Atmos+360°)
- did I say that converting a 64kbps to a 256kbps file would give a better file ? or do I need to understand that ?
Never
, because it would be stupid to think so, and it doesn't prove anything between a 256 and a 512 file (you have to use a higher original one, not a lower one).
So, try to prove what you want, you're free to do it, but please, stop playing with words to make people saying something they didn't ;)

Second, better (and faster in this case) than reading a lot of papers : a test that may not be perfect, but that you could have easily done yourself before assuming other people are wrong (when again, they never said such a thing, because I never did this test before) :

1 - I took the first FLAC I found in my computer, a 24/192 file (not sure if it's the best to test, as/but it adds downsampling in the conversion process). If needed, I can do it again with a 16/44.1 FLAC file.

2 - I used Foobar (it uses iTunes library if I'm not wrong) to convert it to 4 AAC files with different bitrates : VBR320 - VBR256 - CBR512 - CBR256.
I also did VBR in case it'd be needed later, but the last two (in bold) are the one to test as you said there is no difference between AAC 256 and AAC 512, which is certainly right (it may be worst, would need a test to confirm it) if you convert a AAC 256 to a AAC 512 as it won't recover any information lost at first, but may be wrong from FLAC to AAC 512 and AAC 256
So let's see the files : as the original was a 192kHz file, the conversion gave 48kHz AAC files
It appeared that the bitrate of the CBR512 was actually 320kbps, so maybe it doesn't go over 320 even if we select 512.

View attachment 167874

Let's forget about VBR, and just look at the 256 and 512 (which is actually only 320) : DeltaWave difference test :
View attachment 167878

To compare, doing a Digital->Analog then Analog-Digital loopback recording with one of my audio interfaces gives me more than twice a better result than that... ;)
Would it be heard by all people, I don't know and it's not the subject of what you were saying, but regarding the "no difference" like you said, it appears that there are differences (unless I made a big mistake :))
aac_quality.png


...considering the fact that the question has began with my mistake, just let me refer to it first.

I've mentioned a placebo effect and this MP3 to WAV conversion because I've meant the audible difference. That's the purpose of higher quality right? To be able to actually hear it. I could test it with my ears though but I know it wouldn't be any proof for you guys, and while you've mentioned the ABX test which I'm not familiar with yet, it would be good if someone could clear this case up once and for all.
 
Last edited:

Grooved

Addicted to Fun and Learning
Joined
Feb 26, 2021
Messages
679
Likes
441
Hi,

If I understand correctly what was your mistake now, you thought that AAC 320 would be like AAC 256 but just a higher bitrate, without any changes, like if we were converting an MP3 file to Wav. Is that correct ?
I was not getting the MP3 to WAV conversion argument, because everybody should know that you won't get better from up converting anything. The only case you may want to do such a thing would be for format use (for example if you had one device that could read WAV but not MP3).

Back to AAC, in this short exemple to see if they were difference between AAC 256 and AAC 320, I use a "master" to create both files in the same conditions, just the encoder setting being changed.
There were measurement difference, which doesn't tell if it's audible or not.
To test it, you can do a ABX test, which is possible with Foobar after adding the ABX component ( https://www.foobar2000.org/components/view/foo_abx )
You just put both files in Foobar, then select both and right-click, choose "Utilities", then "ABX tracks"

Note that it's not a test that has few chances to succeed with such a small difference of bitrate, but you can try ;)
I did not test it yet, but from past test I did, being tired always give me worst results, and there are sometimes differences that I can hear better with headphones than speakers, and for other files, I can hear it on speakers but not on headphones...
 
Last edited:

abdo123

Master Contributor
Forum Donor
Joined
Nov 15, 2020
Messages
7,425
Likes
7,941
Location
Brussels, Belgium
Hi,

I understand it was a mistake, but the MP3 to wav has nothing to do with that, everybody should now that you won't get better from up converting anything. The only case you may want to do such a thing would be for format use (for example if you had one device that could read WAV but not MP3).

Back to AAC, in this short exemple to see if they were difference between AAC 256 and AAC 320, I use a "master" to create both files in the same conditions, just the encoder setting being changed.
There were measurement difference, which doesn't tell if it's audible or not.
To test it, you can do a ABX test, which is possible with Foobar after adding the ABX component ( https://www.foobar2000.org/components/view/foo_abx )
You just put both files in Foobar, then select both and right-click, choose "Utilities", then "ABX tracks"

Note that it's not a test that has few chances to succeed with such a small difference of bitrate, but you can try ;)

you need to use a song or a sample that is known to break 256 AAC, and check if the added bit-rate mitigate the issue. The current test doesn't really give any new information. 128kbps AAC could be enough for transparency with that song / sample and we wouldn't know.
 

Grooved

Addicted to Fun and Learning
Joined
Feb 26, 2021
Messages
679
Likes
441
you need to use a song or a sample that is known to break 256 AAC, and check if the added bit-rate mitigate the issue. The current test doesn't really give any new information. 128kbps AAC could be enough for transparency with that song / sample and we wouldn't know.
I'm not sure to understand, as I see that you answered on my first post, while I was re-writing it a bit to be more clear.
What song are you talking about ?
And what "new information" ?
I admit that having made this test in another thread to answer to a specific post and starting it again here is making me a bit lost in what we should test or not, and if we are really talking about the same thing :)

I will do the same thing with a 16/44.1 file to see if the difference measured are in the same order
 
Last edited:
OP
GrO

GrO

Member
Joined
Nov 23, 2021
Messages
49
Likes
5
Location
Poland
If I understand correctly what was your mistake now, you thought that AAC 320 would be like AAC 256 but just a higher bitrate, without any changes, like if we were converting an MP3 file to Wav. Is that correct ?

Yup.

I was not getting the MP3 to WAV conversion argument, because everybody should know that you won't get better from up converting anything.

I remember some people who actually didn't and probably there are still some.

The only case you may want to do such a thing would be for format use (for example if you had one device that could read WAV but not MP3).

Yeah I'm aware of that but it's rarely the case these days.
 

arpinnurmela

Active Member
Joined
Jul 12, 2021
Messages
121
Likes
21
Location
Sacramento
I use 320AAC (VBR) for my subsonic server. I'm very happy with the quality. I get most of the detail that's present on my CDs in lossless playback. I've tried 256AAC (VBR and CBR) but have never been excited about how it sounds. I hear too many compression artifacts in the higher frequencies and the low end sounds less full than on 320. I think the differences are subtle but noticeable when using quality in-ears on a quality portable amp. I don't prefer to listen to 320 at all while at home. My wife has Tidal hifi, and we enjoy Tidal for its broad catalog rather than for its SQ. Presently we use Tidal as a music discovery tool such that if we enjoy the music, we just buy the cd or vinyl.
 

ZolaIII

Major Contributor
Joined
Jul 28, 2019
Messages
4,069
Likes
2,409
Every compression format has it's curve and characteristic artifacts. Approach to the problem differs a lot.
AAC is hire for very long time and there are lots of versions and implementations of it ant not rarely it's more to the implementation than codec itself but every little bit counts.
Regarding implementation I highly prize Opus which is actually OGG on high bit rate with fast impulse response.
 

arpinnurmela

Active Member
Joined
Jul 12, 2021
Messages
121
Likes
21
Location
Sacramento
Regarding implementation I highly prize Opus which is actually OGG on high bit rate with fast impulse response.
I'm a big fan of the OGG format as well. And you're right about the version and implementation. I've definitely heard that come into play. But generally I'd say that MP3, to me, sounds inferior to similar bit rate AAC and OGG, but that the difference is subtle and could just be in my mind.
 

ZolaIII

Major Contributor
Joined
Jul 28, 2019
Messages
4,069
Likes
2,409
I'm a big fan of the OGG format as well. And you're right about the version and implementation. I've definitely heard that come into play. But generally I'd say that MP3, to me, sounds inferior to similar bit rate AAC and OGG, but that the difference is subtle and could just be in my mind.
Depends on the curve, actually MP3 at 320 beat them but let's say 196 Kb MP3 is similar to 160 Kb AAC - 176 OGG.
Opus uses impulse response and 32 bit FP precision to mask the artifacts sort of feeling it more natural while codec is the same (container differs).
Actually didn't give it up a look (@ Opus development) from multichannel audio update.
 

arpinnurmela

Active Member
Joined
Jul 12, 2021
Messages
121
Likes
21
Location
Sacramento
I'm not doubting your statement per se, I'm just wondering how you know that MP3 beats out AAC and OGG? I've tried LAME for MP3 but even with LAME I felt that AAC had a slight edge. And if not LAME, then which MP3 compression does beat out AAC and OGG.
 

ZolaIII

Major Contributor
Joined
Jul 28, 2019
Messages
4,069
Likes
2,409
I'm not doubting your statement per se, I'm just wondering how you know that MP3 beats out AAC and OGG? I've tried LAME for MP3 but even with LAME I felt that AAC had a slight edge. And if not LAME, then which MP3 compression does beat out AAC and OGG.
From a bunch of papers (white or otherwise). That would be at 98% of fidelity to the original as I recall. Fidelity is not per se goal of compression codecs but acceptable quality at higher compression rates. Things change drastically on low bit rates. Opus again beats the mentioned bunch on low bit rates (where it used GSM suitable codec) and in mid one's (where it uses both codes and all response precisions [8 to 32 bit]). Better AAC implementations are not far behind. For comparison purpose only Opus @ 64 ~AAC @ 96 ~ MP3 @ 128.

Edit: forgot to mention how MP3 also had a long history of revisions and how there ware more advanced implementation that never got a wider adoption (Farens Thompson MP3Pro and later surround versions for instance).
 
Last edited:

bravomail

Addicted to Fun and Learning
Joined
Oct 19, 2018
Messages
817
Likes
461
I tested FLAC file on song "Hush" by Deep Purple. My rating: OGG 320< MP3 CBR 320< AAC VBR 320.
It is heavily distorted track and even with AAC I noticed issues.
U can conduct ur own tests - Foobar2000 can do encoding (for AAC codec - install iTunes, foobar will pick it up). Select a demanding track - my choice is heavy metal distorted guitars.
 

ZolaIII

Major Contributor
Joined
Jul 28, 2019
Messages
4,069
Likes
2,409
Out of curiosity just did a conversation of good quality jazz track (96 KHz 24 bit flac) and somewre at the middle of it thought what if I whose doing it wrong.
Did a quick conversation trough DbPowerAmp (for a sakes of integrated sampling rate convertor or better say to minimise it's impact) on single core and at first tried to trigger present supported lossy codecs at their best. Worst whose (by all means) Lame MP3 stereo ABR, then AAC He V2 SBR PS (boxi and lose in highs), Opus hybrid high bit rate mode (both codecs and variable FP used which whose less boxi but still highs ware not all that convincing) all approximately around 276~288 Kb/S. They all did a bandwidth cut (what's once lost is permanently lost) no matter what the tag says. MP3 did to CD quality (9.45 MB size to original 46.86 MB 16 bit 80% compression), AAC and Opus did to 96 KHz 16 bit 102.02 MB(Opus 9.31 MB 91% compression rate [encoding is actually 2x 20 KHz], AAC 9.69 MB 91%).
Than I thought what a hell let's try it from another perspective. Lossy high fidelity oriented codec or better say lossy portion of WavPack. So I did Hybrid mode one resulting with 439 Kbps 14.61 MB and a no cut in bandwidth 96 KHz 24 bit - 153.02 MB 91% compression rate. Timing whose perfectly smooth but highs still need some energy focusing from time to time. So I went to lossy + lossless (correction graph) and raised a target to 620 kbps resultsing in 20.61 MB size to original and compression rate of 87%. This one whose undistinguishable to original for me. Of course its all subjective and I used my ears as well as my eye sight (response graphs).
For me conclusion is that there are valuable lossy formats worth considering as alternative to compressed wav's (flac or Apple losseles) and that the almost 5x size difference with no losses of original bandwidth or structure is still worth it it remains to be seen if it gets wider support in the future.
PS: why the hell did I do this in the first place?
 
Last edited:
OP
GrO

GrO

Member
Joined
Nov 23, 2021
Messages
49
Likes
5
Location
Poland
Thank you for your testings guys now the topic has been cleared up as for AAC 256 kbps and 320 kbps, but can we afford to say the same comparing 320 and 512 kbps, and if 512 is even worth using regarding the file size to the quality benefits ratio if there are any?

Presently we use Tidal as a music discovery tool such that if we enjoy the music, we just buy the cd or vinyl.

Yeah, that's also the only reason for using such services by me either considering the music, but I also remember the times when internet was not so popular and services like 'youtube', 'spotify', or 'tidal' didn't even exist. When you knew one or two songs per album maybe from MTV, and then you bought an original one where there was some magic related while discovering tracks you've never heard before, even if you didn't instantly like them at the first glance.

Opus uses impulse response and 32 bit FP precision to mask the artifacts

By "FP" you mean the floating-point filter?
 

arpinnurmela

Active Member
Joined
Jul 12, 2021
Messages
121
Likes
21
Location
Sacramento
Thank you for your testings guys now the topic has been cleared up as for AAC 256 kbps and 320 kbps, but can we afford to say the same comparing 320 and 512 kbps, and if 512 is even worth using regarding the file size to the quality benefits ratio if there are any?



Yeah, that's also the only reason for using such services by me either considering the music, but I also remember the times when internet was not so popular and services like 'youtube', 'spotify', or 'tidal' didn't even exist. When you knew one or two songs per album maybe from MTV, and then you bought an original one where there was some magic related while discovering tracks you've never heard before, even if you didn't instantly like them at the first glance.



By "FP" you mean the floating-point filter?
Yep... MTV. The old days of a thing called television.

I just miss all of the record stores that used to exist seemingly everywhere. Now I have to drive 30 minutes to find an okay one.
 
  • Like
Reactions: GrO

ZolaIII

Major Contributor
Joined
Jul 28, 2019
Messages
4,069
Likes
2,409
@GrO by FP I mean floating point (x.xx) precision opposite to full numbers only.
As they all (AAC, MP3 and Opus) made a bandwidth cut and permanent change to 96 KHz 24 bit source I can't really recommend any of them (tho AAC and Opus supposedly only did 24 bit too 16 bit [in reality they did lose much more]). Tho I don't have much of (raal) such sources I expect I will have more and still won't recommend those for 16 bit one's. Can't really recommend WavPack either until support for it grow.
 
Last edited:
  • Like
Reactions: GrO
OP
GrO

GrO

Member
Joined
Nov 23, 2021
Messages
49
Likes
5
Location
Poland
They all did a bandwidth cut (what's once lost is permanently lost) no matter what the tag says.

Yeah, once I've realized it back in the days I've just dropped all lossy-compression-codecs at once.

Yep... MTV. The old days of a thing called television.

I just miss all of the record stores that used to exist seemingly everywhere. Now I have to drive 30 minutes to find an okay one.

Yeah, same thing here about all those music shops and many times it's just easier to order via internet, and considering a television I'm very happy I've dropped it too for many different reasons.

@GrO by FP I mean floating point (x.xx) precision opposite to full numbers only.

Thanks for clearing it up.
 

Grooved

Addicted to Fun and Learning
Joined
Feb 26, 2021
Messages
679
Likes
441
Thank you for your testings guys now the topic has been cleared up as for AAC 256 kbps and 320 kbps, but can we afford to say the same comparing 320 and 512 kbps, and if 512 is even worth using regarding the file size to the quality benefits ratio if there are any?

That's a good point. I didn't check it before, but just checked a same track with AAC 320 and FLAC :
- AAC 320 : 13.8MB
- FLAC 16/44.1 : 21MB
It's only 50% bigger, and as AAC 512 would be bigger than AAC 320, I'm really not sure why we would need AAC 512.
 
  • Like
Reactions: GrO
D

Deleted member 23982

Guest
- did I say that converting a 64kbps to a 256kbps file would give a better file ? or do I need to understand that ?
Never
, because it would be stupid to think so, and it doesn't prove anything between a 256 and a 512 file

i cant agree with this, ever heared of upsampling? imagine having two datapoints one at 0% volume, the other at 100% volume, now imagine doubling sampling size and having a third datapoint with 50% in between the two first ones, thats why upsampling actually makes a difference to the better (tho its a simplificated way to explain it)

thats why high khz files exist, its not about hearing above 20khz but its about having more datapoints so fast volume changes are more fluid and therefore sound better
its jsut that 192 khz files are named that way because in "theory" they can hold data that reproduces 192khz (or half that because of two channels) with the datapoints in the file, for 44,1khz material it means more resolution in the datapoints

sorry for my bad english, but i hope you could understand what i tried to say :)
 
Last edited by a moderator:

BDWoody

Chief Cat Herder
Moderator
Forum Donor
Joined
Jan 9, 2019
Messages
6,948
Likes
22,625
Location
Mid-Atlantic, USA. (Maryland)
thats why high khz files exist, its not about hearing above 20khz but its about having more datapoints so fast volume changes are more fluid and therefore sound better
its jsut that 192 khz files are named that way because in "theory" they can hold data that reproduces 192khz (or half that because of two channels) with the datapoints in the file, for 44,1khz material it means more resolution in the datapoints

That's a bunch of nonsense.

Maybe a little Monty is in order here...

 
Top Bottom