• 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!

Formats: 320k aac vs Lossless (Roon related)

Dj7675

Major Contributor
Forum Donor
Joined
Jan 12, 2019
Messages
2,116
Likes
2,781
I am in a bit of a quandary. Our only home broadband option is a DSL 3M speed and I want to use Roon + Tidal. If there is no internet activiting I can stream lossless from the Tidal app (or volumio via shd). However, when I signed up for the Roon trial (which I really like), even with no other internet activity, Roon/Tidal won't work well enough.
My only real option would be to downgrade Tidal to premium which streams at 320k AAC if I wanted to stream with Roon.
From reading here, I have seen comments that 256K and up it may be difficult to tell the difference betweek 256K aac and lossless. Has anyone done any DBT using 320k aac / Lossless and could you tell the difference? I would like to do a double blind test of 320k aac and lossless. Can anyone point me to a test anywhere testing these? Or how to set it up on my own pc? My thought is if I can't tell the difference I would just downgrade to the Tidal Premium 320k so I can use Roon.
What I found while using Roon, is it was helpful in discovering new music, so if there isn't an audible difference between 320k aac and lossless, that is the direction I will go.
Appreciate any insights.
 

Soniclife

Major Contributor
Forum Donor
Joined
Apr 13, 2017
Messages
4,500
Likes
5,417
Location
UK
You can convert a flac you own to aac and then try them in a blind test tool, e.g. foobar.
Different aac encoders produce different results, ideally you would find out what tidal use and also use that, but that might be tricky to find out.

If tidal use a good aac encoder it will be tricky to impossible to tell from flac.

Honestly if the flac tier isn't stable I'd just downgrade and enjoy the music.
 
OP
D

Dj7675

Major Contributor
Forum Donor
Joined
Jan 12, 2019
Messages
2,116
Likes
2,781
You can convert a flac you own to aac and then try them in a blind test tool, e.g. foobar.
Different aac encoders produce different results, ideally you would find out what tidal use and also use that, but that might be tricky to find out.

If tidal use a good aac encoder it will be tricky to impossible to tell from flac.

Honestly if the flac tier isn't stable I'd just downgrade and enjoy the music.
Appreciate your reply. I think that is what I am going to do. I have not ever used foobar, but I will start to do some research. I might start by ripping a some tracks to aac / flac and see if I can set up a DBT to see if it is audible to me.
 
OP
D

Dj7675

Major Contributor
Forum Donor
Joined
Jan 12, 2019
Messages
2,116
Likes
2,781
As noted briefly while I was posing this, download Foobar2000, download the Foobar encoder pack, download the Foobar A/B/X comparometer, give yourself an afternoon or two or three to figure out how the heck to set it all up (it works! I swear!), rip a file from a CD you think is pretty awesome sound wise to lossless and then to 256 AAC or really whatever codec and bitrate you want to try, and have your hand at a blind level matched a/b/x test. If you can tell the difference between 256 AAC and lossless please post your results (the comparometer will spit out a nice chart for you) because I will be mightily impressed.

Once you play around with the ABX stuff you will get the big picture. I won’t spoil it for you. ; )

To me Tidal lossless tier is pointless. It’s nice to figure it out for yourself as above but @Soniclife ‘s suggestion is very sound based on the experience of people who have done the footwork of a/b/x testing for themselves—just downgrade and enjoy the music.
Thank you for further info to get me started. Should be fun and interesting. I was surprised that Tidal offered 320k aac. In the past I have had apple music which does 256k aac and thought it sounded good to me but have not ever compared them to lossless like this before.
 
OP
D

Dj7675

Major Contributor
Forum Donor
Joined
Jan 12, 2019
Messages
2,116
Likes
2,781
Are there any particular styles/genre's of music which are more susceptible to compression?
 

Soniclife

Major Contributor
Forum Donor
Joined
Apr 13, 2017
Messages
4,500
Likes
5,417
Location
UK
Are there any particular styles/genre's of music which are more susceptible to compression?
Go for complex wide bandwidth music with a lot going on, I find big studio rock stuff that's packed with things going on, and lots of studio processing etc. I'm not sure if there is any science to guide you, it might just depend on the listener.
 

Soniclife

Major Contributor
Forum Donor
Joined
Apr 13, 2017
Messages
4,500
Likes
5,417
Location
UK
I was surprised that Tidal offered 320k aac. In the past I have had apple music which does 256k aac and thought it sounded good to me but have not ever compared them to lossless like this before.
Apple used 256 for iTunes downloads, in a vbr setting that was very good, probably better than 320 fixed, if they are still using that it could be better than tidal at lower rate, but too complicated to guess. I expect the difference is minimal, and would not guess based on simple numbers which is superior.
 

Soniclife

Major Contributor
Forum Donor
Joined
Apr 13, 2017
Messages
4,500
Likes
5,417
Location
UK
To me the Tidal lossless tier is pointless.
To me it's worth it for the simple avoidance of doubt, I would drive myself mad with trying to get to the truth, but I'm not advocating this as a sensible approach. I do very happily use Spotify when it's not on tidal, and I've never thought it worse.
 

flipflop

Addicted to Fun and Learning
Joined
Feb 22, 2018
Messages
927
Likes
1,240
Properly encoded 256 kbps and 320 kbps AAC will both sound transparent for all intents and purposes.

If you're still having trouble setting up a DBT, please have a look at this thread: https://www.audiosciencereview.com/forum/index.php?threads/codec-abx-testing.7524/
Everything should be covered by it. Let me know if anything needs to be clarified.

As for the music that's the most revealing of compression artifacts, see this thread: https://www.audiosciencereview.com/...sic-tracks-for-speaker-and-room-eq-testing.6/

Looking forward to seeing your report log :)
 

Tks

Major Contributor
Joined
Apr 1, 2019
Messages
3,221
Likes
5,494
Listening to some modern music, I can't even tell the difference between ~200kbps VBR and FLAC..

I mean I can, but nothing approaching confidence ratings of 95%+

On some days I get 80%'s and some days sub 50%'s, which is essentially statistically insignificant.
 
OP
D

Dj7675

Major Contributor
Forum Donor
Joined
Jan 12, 2019
Messages
2,116
Likes
2,781
This is a pretty enlightening process. Before going to the trouble of setting this up on my own I decided to first find a few online tests. I have just completed the first one at the link below. Looks like they have several tests to choose from 48% on the first test. Pretty familiar with their samples too. Had no idea it was that difficult to do. Like most things I’m sure most have seen the tests at the link below but in case not it was very interesting to me. Quite a few different tests there to do. Without having a great ear for music, or training very challenging!
http://abx.digitalfeed.net/list.html
Edit: Funny thing is after the test I needed to listen to some Daft Punk.
 
  • Like
Reactions: Tks

fulffy512

Member
Joined
Jun 1, 2019
Messages
46
Likes
21
The issue with AAC/Ogg is that for some dark ambient and noise industrial they can't handle it without sounding like a bad mp3, yet with lame in the 192 - 320k area no issues beyond 4 tracks that can fixed by using opus/musepack on them instead.

Opus is getting better since the Vorbis codec is dead and the lead dev for the 3rd party encoder hasn't updated it at all. There allot of issues that been ignored like just upping the bit rate instead of using other compression tricks that vorbis has.

Just shows how far a fully maxed out encoder can do for flawed codecs like MP3.
 
OP
D

Dj7675

Major Contributor
Forum Donor
Joined
Jan 12, 2019
Messages
2,116
Likes
2,781
I will continue to do some testing with lower bitrates vis dbt tests to see where I have some degree of certainty that I can tell the difference. So far, 320k aac vs lossless, I cannot. It would be good to know what that threshold is. If I were guessing I would say 256k aac or similiar. I do remember ages ago mp3 128k did not sound very good compared to 192-256k. But I never did anything like this testing before.
 

Jim777

Active Member
Forum Donor
Joined
May 28, 2019
Messages
124
Likes
203
Location
Greater Boston
A critical one is stereo applause, if that sounds the same, the time smearing is not an issue. My guess is that you'll be perfectly fine with such bit rates.
 

fulffy512

Member
Joined
Jun 1, 2019
Messages
46
Likes
21
And from more testing ~128kbps Opus sounds 99% for all my music and is on par/better than the ones in the 256k area. Only 1 album needs like 400kbps, its a harsh noise wall album if your wondering what it is.

Edit: lol Ogg Vorbis sounds better at 160Kbps after more messing around, Opus VBR is way to constrained and the whole resample to 48Khz not being perfect.


Are there any particular styles/genre's of music which are more susceptible to compression?

Electro and experimental is the biggest pain for any lossy codec, With so many artificial sounds and noises that its common for Lame MP3 to need 320kbps and AAC/Vorbis/OPUS = 180 ~ 500Kbps. Hence why i sometimes find the whole "320kbps is enough" claim to be annoying or red flag there tastes might be narrow. Since artists like Merzbow and others have content that sounds like crap unless you abuse the 512kb/s setting newer codecs have. lol
 
Last edited:

daftcombo

Major Contributor
Forum Donor
Joined
Feb 5, 2019
Messages
3,687
Likes
4,068
The issue with AAC/Ogg is that for some dark ambient and noise industrial they can't handle it without sounding like a bad mp3
What kind of sounds in that kind of music would be responsible for such?
 

q3cpma

Major Contributor
Joined
May 22, 2019
Messages
3,060
Likes
4,416
Location
France
What kind of sounds in that kind of music would be responsible for such?
I guess he means music with a lot of noise. Compressing random noise is always hard (as in impossible because entropy is maximum), clapping being an example that defeats most lossy codecs.
But to be honest, I listen to very raw black metal with plenty of noise and I use Vorbis (aotuv) q5 with no problem at all.

As a bit of fun, I tried to characterize what style would make Vorbis overshoot the most (the target being 160 kbps):
Code:
$ find . -type f -name '*.ogg' | while IFS= read -r i; do printf '%g %s\n' "$(ogginfo "$i" | sed -n '/Average bitrate/s#.*Average bitrate: \([^ ]*\) .*#\1#p')" "$(printf '%s' "$i" | cut -d/ -f3-)"; done | sort -t1 -g | tee bitrate.log | tail -n30
198.008 Johann Sebastian Bach/BWV 827_ Partita No. 3 in A minor/03. III.ogg
198.081 Castlevania/(1991-1) Castlevania II_ Belmont's Revenge (Game Boy)/19. The End Of the Day (Staff Roll BGM).ogg
198.262 Johann Sebastian Bach/BWV 826_ Partita No. 2 in C minor/05. V.ogg
198.823 Johann Sebastian Bach/BWV 829_ Partita No. 5 in G major/07. VII.ogg
198.848 Psychotic Waltz/(1996) Bleeding/06. Drift.ogg
198.995 Johann Sebastian Bach/BWV 827_ Partita No. 3 in A minor/05. V.ogg
199.784 Johann Sebastian Bach/BWV 825_ Partita No. 1 in B♭ major/02. II.ogg
200.089 Castlevania/(1991-1) Castlevania II_ Belmont's Revenge (Game Boy)/18. Union (Ending BGM).ogg
200.222 Johann Sebastian Bach/BWV 827_ Partita No. 3 in A minor/04. IV.ogg
200.518 Johann Sebastian Bach/BWV 828_ Partita No. 4 in D major/07. VII.ogg
200.674 In Flames/(1995) Subterranean/04. Timeless.ogg
200.936 Johann Sebastian Bach/BWV 828_ Partita No. 4 in D major/06. VI.ogg
201.03 Johann Sebastian Bach/BWV 827_ Partita No. 3 in A minor/01. I.ogg
201.506 Johann Sebastian Bach/BWV 829_ Partita No. 5 in G major/03. III.ogg
202.104 Yes/(1971-1) The Yes Album/02. Clap (studio version).ogg
202.734 Johann Sebastian Bach/BWV 829_ Partita No. 5 in G major/02. II.ogg
203.577 Le Orme/(1974) Contrappunti/04. India.ogg
203.634 Johann Sebastian Bach/BWV 829_ Partita No. 5 in G major/05. V.ogg
203.979 Johann Sebastian Bach/BWV 825_ Partita No. 1 in B♭ major/05. V.ogg
204.014 Castlevania/(1991-1) Castlevania II_ Belmont's Revenge (Game Boy)/04. Road of Enermy #1 (Before Boss BGM).ogg
206.713 Bathory/(1988) Blood Fire Death/08. Blood Fire Death.ogg
207.843 Lucifugum/(2001) …а колесо всё cкрипит… (…and the Wheel Keeps Crunching…)/07. Under Rustle of Branchs [Bonus].ogg
208.791 Nokturnal Mortum/(2004) Мировоззрение/05. Печаль родных земель.ogg
209.115 Psychotic Waltz/(1996) Bleeding/09. My Grave.ogg
212.147 David Bowie/(1970) The Man Who Sold the World/08. The Man Who Sold the World.ogg
213.688 Pigface/(1991) Gub/10. War ich nicht immer ein guter Junge_ War ich nicht immer schön und nett_ Ich zerpflückte niemals eine Spinne - war niemals frech und stahl.ogg
215.131 Castlevania/(1991-1) Castlevania II_ Belmont's Revenge (Game Boy)/08. Ripe Seeds (Plant Castle BGM).ogg
217.149 Castlevania/(2009) Castlevania_ The Arcade (Arcade)/07. Poison Mind ~Mini-Boss Theme~ (From “Akumajo Dracula”).ogg
230.021 The Neon Judgement/(2010) Early Tapes/12. Concrete (Vegas) [Live at Arnoz].ogg
241.545 The Neon Judgement/(2010) Early Tapes/01. Factory Walk.ogg
Can't find the common denominator, to be honest. Looks like string instruments with high frequencies kept is a theme.
 
Last edited:

John Dyson

Active Member
Joined
Feb 18, 2020
Messages
172
Likes
90
Chiming in -- my project attempts to do improved processing of audio, and i have done careful comparisons between mp3 and raw audio... My hearing is limited as I am 63yrs old and hear nothing above about 14kHz -- but bandwidth isn't the big deal about mp3. On my A/B testing, the biggest impairment that I can hear -- temporal resolution. Repetitive quick little impulses of a certain kind can be lost or attenuated in mp3 even at the highest rate. Opus does it better, but the result is more of a random lossage with more remaining detail rather than smooth lossage on mp3.

These are a result of careful testing with the goal of precision and not 'sounding good'. I wasn't focused on sounding good or bad at the time, but only precision and details lost or synthesized.

For simple, casual listening -- not burying my ears into headphones -- max rate mp3, high rate opus or raw audio -- I cannot casually tell the difference.

* Clarifying my observation about loss of temporal resolution -- it IS there, there IS a loss, but I could only hear it in limited circumstances. I do hours and hours and hours of critical, NOT PLEASURE listening, and the differences might not be audible to many people. The differences are very slight. After doing some re-reseach about Opus and MP3, my observation about some loss of temporal resolution (and it is VERY slight, and never strongly manifested) is as expected per the design and mathematics in the schemes.


John
 
Last edited:

sergeauckland

Major Contributor
Forum Donor
Joined
Mar 16, 2016
Messages
3,440
Likes
9,100
Location
Suffolk UK
I've done a fair few blind tests on digital vs analogue and uncompressed vs compressed dgital audio over the years.

Whilst I can tell, just about, the difference between uncompressed and compressed in an AB comparison, (obviously depending on bit rate) I find it impossible to tell, just from listening to a piece of audio, whether it's uncompressed or compressed without the AB comparison. I've even gone down to very low bit rates, like 64k, and in the absence of anything to compare it with, it's been OK. I wasn't immediately aware of anything being amiss, albeit with 64k, that was listening on 'office' 'speakers, not my main system.

Even on my main system, I find 128k MP3 perfectly acceptable on internet radio stations like Venice Classical or Swiss Jazz, but clearly, I don't have the original to compare with.

With BBC Radio at 320kbps AAC, I find it better than FM, at least lower noise and wider bandwidth (not that I can now hear above 15k) whilst Radio Paradise, I can't hear any difference whatsoever between their 320k stream and their FLAC stream.

My conclusion is that yes, with critical listening and an AB comparison, it may be possible for some to hear the difference between compressed and uncompressed, although I question whether anyone can hear that at 320k, and then only with an AB comparison. Just listening to a stream without knowing what it is and with no comparison, pretty much all will sound fine.

S.
 
Top Bottom