• 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!

Songs that sound better in in lossless vs compressed streaming

Chrispy

Master Contributor
Forum Donor
Joined
Feb 7, 2020
Messages
7,757
Likes
5,916
Location
PNW
The idea that the streaming service actually remasters thousands of recordings....nah.
 
OP
R

RC231

Member
Joined
Jul 31, 2020
Messages
41
Likes
21
Sounds like most of the difference people are hearing is due to different masters and I am chasing minuscule gains. I will keep my amazon trial to play with to see if there are any exclusive artists and likely sign up for multi account spotify when it is up. Spotify has 2 accounts for $12.99 or up to 6 for $14.99 both are reasonable values.
 

krabapple

Major Contributor
Forum Donor
Joined
Apr 15, 2016
Messages
3,169
Likes
3,717
From the point of view of Audio SCIENCE Review, Lossy is lossy and lossless is lossless, this is not an apples to apples comparison and how things sound is subjective and not scientific, or measurable in a meaningful way.

Where on earth did you get that idea? Not at ASR and not from SCIENCE.

If you are removing data, it is not the same as lossless and compressed.

I think he knows that. But being 'not the same' doesn't necessarily mean 'sounds different'.

There is no inherent flaw in comparing lossy to lossless audio. The problems with comparing in this user's case is that the user can't do volume matching, and can't know of the mastering is the same on the two sources.
 
Last edited:

krabapple

Major Contributor
Forum Donor
Joined
Apr 15, 2016
Messages
3,169
Likes
3,717
Not all MQA-encoded music is remastered and I don't think this is a requirement, but some albums are for some reason, maybe because the song needs a remaster or they just want to provide more value with your acquisition. An example of this is the album "Black Widow" by In this Moment. When I was a Tidal user, the MQA version had the cymbals better defined, and this is something I didn't need a double-blind to notice. I remember I was switching between the regular version and the MQA quickly, and I clearly could hear the difference in that album. I think there are other examples out there, but a remaster for MQA wouldn't be a strange thing to do.

Maybe Tidal/MGA does processing to its stream (e.g. dynamic range compression, EQ) that your other source doesn't, or vice versa . Maybe it's just a level difference. Maybe it's different source tapes.. Maybe it's different mastering. You can't know without a more in depth look.

I know that on HDtracks (high rez downloads), simply choosing the 196/24 gets you a *different mastering* of some albums* versus 96/24 or 44/16 . This is verifiable from examining the waveforms, and listening. It's also annoying, because it is not advertised.


*the Van Halen catalog, for one -- the highest rez version is free of the dynamic range compression of the other masterings
 

UCrazyKid

Active Member
Joined
Nov 15, 2019
Messages
185
Likes
238
Location
NW Chicago Suburbs
Where on earth did you get that idea? Not at ASR and not from SCIENCE.



I think he knows that. But being 'not the same' doesn't necessarily mean 'sounds different'.

There is no inherent flaw in comparing lossy to lossless audio. The problems with comparing in this user's case is that the user can't do volume matching, and can't know of the mastering is the same on the two sources.
Two things I would like to mention. First, my comment is based on what the OP asked BEFORE he edited this question and the thread title. So it isn’t exactly relevant any more. Second, are you suggesting that lossy is equal to lossless and that there is no science to the contrary? I’m confused by your comment and your aggressive and demeaning attitude.
 
OP
R

RC231

Member
Joined
Jul 31, 2020
Messages
41
Likes
21
I only edited my title as the original post was always spotify vs amazon HD.
I would be very interested in measurements showing the differences between the reconstructed signals between different amounts of lossy compression vs lossless but I don't know how you would compare them in an objective way.
 

krabapple

Major Contributor
Forum Donor
Joined
Apr 15, 2016
Messages
3,169
Likes
3,717
Two things I would like to mention. First, my comment is based on what the OP asked BEFORE he edited this question and the thread title. So it isn’t exactly relevant any more. Second, are you suggesting that lossy is equal to lossless and that there is no science to the contrary? I’m confused by your comment and your aggressive and demeaning attitude.


Too bad. I am 'suggesting' that lossy and lossless may be *audibly* indistinguishable in many cases, which is true as per SCIENCE. Lossy algorithms after all, were *designed* by engineers based on psychoacoustic research, to be as transparent as possible but also allowed for lower-quality implementations that were more likely to be audible.
 

krabapple

Major Contributor
Forum Donor
Joined
Apr 15, 2016
Messages
3,169
Likes
3,717
I only edited my title as the original post was always spotify vs amazon HD.
I would be very interested in measurements showing the differences between the reconstructed signals between different amounts of lossy compression vs lossless but I don't know how you would compare them in an objective way.

The important thing is whether the measurable differences are audible. Many a shenanigans in audio have arisen from showing a lossy vs lossless waveform to frighten audiophiles. They will surely look different. But difference is not always audible.
 
Last edited:

tmtomh

Major Contributor
Forum Donor
Joined
Aug 14, 2018
Messages
2,636
Likes
7,497
I prefer lossless and will seek it out over even high-bitrate lossy (e.g. 256k AAC or 320k mp3) whenever possible.

However, I prefer the idea of lossless and I feel better about it. I do not actually believe that I could reliably distinguish lossless from high-bitrate lossy in a blind test - I would guess that the best I could do would be to tell the difference reliably with some kinds/passages of music but not with most music. And in fact I have my doubts that I would easily or reliably distinguish the two in a sighted test either.

If someone told me that lossless and properly encoded high-bitrate lossy are perceptually 100% identical to 99+% of humans, I would not take that claim at face value, but I would certainly not dismiss it either.
 

NTomokawa

Addicted to Fun and Learning
Forum Donor
Joined
Jan 14, 2019
Messages
779
Likes
1,334
Location
Canada
If someone told me that lossless and properly encoded high-bitrate lossy are perceptually 100% identical to 99+% of humans, I would not take that claim at face value, but I would certainly not dismiss it either.
I mean there was a time when lossy encoders all had their own issues, but it's no longer that time...

Try converting some FLAC files of yours to MP3 using (for example) LAME v0 and ABX them.
 

tmtomh

Major Contributor
Forum Donor
Joined
Aug 14, 2018
Messages
2,636
Likes
7,497
A defensible advantage of lossless is that it is 'archival' -- you can re-encode with the fewest artifacts. Lossy to lossy on the other hand is not recommended.

Excellent point - and one more reason I seek out lossless (and will not purchase lossy unless there is no other choice).
 

luft262

Senior Member
Joined
Mar 25, 2021
Messages
465
Likes
236
Location
Phoenix
Studies have definitely shown that it's possible to discriminate lossily compressed recordings from their lossless originals, although at 320kbps it tends to be very difficult, and to take some training, the right source material, and a good system. So I wouldn't expect there to be any audible difference between the two for most recordings/systems/listeners, although subtle differences may be apparent in some circumstances.

Also keep in mind that a lot of streamed content from major labels appears to be watermarked, which may result in audible quality degradation regardless whether the content is streamed at lossy 16/44 or lossless 24/96.

Personally, as I started with Spotify before a lossless hi-res service with a significant collection of music existed, I have decided to stick with it. I guess there might be some subtle audible differences, but having done a number of 320kbps to lossless blind tests on myself, I've concluded that in most cases I can't hear differences, and in those few cases where I might be able to, the effects of lossy compression are very subtle and don't hinder my enjoyment of the music, anyway.

I think you hit it on the button! Great question/thread and great answer. To add to your response, my wife and I did a blind test where one person would switch between YouTube Premium (320Kbs MP3 or 256Kbs AAC) and Amazon HD Music (lossless/FLAC/etc.) while the other person listened but couldn't see what was being played and after about a dozen songs we each picked YouTube Music about 50% of the time and Amazon the other 50% as what sounded best. We were using HiFiMan Sundara headphones connected to a JDS Atom Amp and JDS Atom DAC. In my opinion that is a decent set of equipment and we listen to a lot of music and we still weren't skilled enough to pick the higher quality music consistently. I think for the vast majority of users/applications AAC 256 and MP3 320 is all you need. CD/FLAC would be technically better from a science/numbers perspective, but rarely if ever would the end user be able to tell. Going beyond CD quality (44.1/16bit) is almost certainly overkill and not worth it for 99.9% of users/scenarios. Hi-Res music is mostly a gimmick IMHO. Unless you are trained to pick out the most minute of details, you enjoy doing so, and you have superb equipment, Hi-Res audio will not be worth it. MQA could offer some quality improvements, but not because it offers much technical quality over AAC/MP3, but because it may have been remastered. Remember, MQA is lossy, so FLAC or a CD is still technically better assuming they are coming from the same master. As long as the streaming service you are using is at AAC 256 or MP3 320 the sound quality should be very good and you can make your decision of which service to use based on content, price, and interface.
 

Killy

Member
Joined
Feb 27, 2021
Messages
10
Likes
44
Location
United Kingdom
I prefer lossless and will seek it out over even high-bitrate lossy (e.g. 256k AAC or 320k mp3) whenever possible.

However, I prefer the idea of lossless and I feel better about it. I do not actually believe that I could reliably distinguish lossless from high-bitrate lossy in a blind test - I would guess that the best I could do would be to tell the difference reliably with some kinds/passages of music but not with most music. And in fact I have my doubts that I would easily or reliably distinguish the two in a sighted test either.

If someone told me that lossless and properly encoded high-bitrate lossy are perceptually 100% identical to 99+% of humans, I would not take that claim at face value, but I would certainly not dismiss it either.

Spot on with the idea of lossless I’m fretting with the complexity of my system re using Roon to dlna hi res to my modest KEF LSXs and HD6XX Topping E30/A50s when I know I can’t hear any difference but can’t help but feel I’m missing out on something. This gets me in a position where I’m considering Roon or Audirvana plus Qobuz when I get Apple Music bundled with my mobile. £300 per annum would be saved by not having a High Res system that can pay for real noticeable increases in fidelity over the years. I think this is the state that many audiophiles get trapped into including myself, I did have a £30k system till I moved. You keep going into the local audio dealer and demoed the latest cables, power supply and racks luckily I just couldn‘t hear any difference. It does though create a nagging itch.
 

sgent

Member
Joined
Mar 24, 2020
Messages
99
Likes
127
I haven't done this since ripping CD's decades ago (when storage was the real issue), but on blind testing I could pick the lossless recording over 90% in songs with a high dynamic range -- and with much of modern music being dynamically compressed there isn't a noticeable difference. Amir also has a video where he uses blind A/B testing and is almost inerrant.

I found the biggest difference comes from clipping stray notes which are dissonant. Try the intro to Dire Straits "Money for Nothing", Brittany Spears "Toxic", Stevie Wonder "Sir Duke" (or anything off that album). Also Jazz and Classical music are much easier to pick out lossy vs lossless because you have multiple instruments playing different tones at different frequencies which is difficult for lossy compression.

All of that said, its easier to pick out something like 128k MP3 vs FLAC, and once you learn what to listen for it becomes easier for 320 vs FLAC.
 

krabapple

Major Contributor
Forum Donor
Joined
Apr 15, 2016
Messages
3,169
Likes
3,717
Actually symphonic classical isn't generally hard to lossy encode. Dynamic range is not the main factor either. Noisy high frequency content can be.

'Killer' samples -- samples that used to test lossy encoder settings, far from typically being multiple instruments or complex, have typically been things like solo harpsichord, electronic music, castanets, industrial metal, certain solo vocals. For examples . Also a discussion of what kinds of sounds are hard to encode well.

The 'tricks' Amir uses to ABX high quality mp3 from source have been discussed. They bear little resemblance to the methods and claims of audiophiles who report an absurd ability to perform nearly instant, obvious discrimination of same.
 
Last edited:
Top Bottom