Head-Fi.org › Forums › Equipment Forums › Sound Science › Lossless vs. Lossy
New Posts  All Forums:Forum Nav:

Lossless vs. Lossy - Page 2

post #16 of 76
Proton007, who mentioned detail? I assume your post is just taking the p#ss and English is not your first language
post #17 of 76
Quote:
Originally Posted by KT66 View Post


i initially ripped all my cds to mp4/320 using Sony Media Go, never thinking i would get into this head fi thing.

I have been a 2 channel audiophile for 25 years, worked in the industry for a while, and have been a musician for longer, spending a lot of time in recording studios and am a musician first and audiophile second.
my 2 channel system got to a point where i couldn't afford to improve or justify any upgrade(eg Radikal and Keel),plus I was and am extremely happy with my system.

So thanks to accidently getting a great audio phone (n8) i did get into head fi and serious listening on the move.
Then i started to re rip to wav or flac using EAC.However there are many CDs still not done.

i did not do abx testing, i did not do blind testing, and my post was not meant to wind anyone up.
I respect that we all hear differently, and some may not hear any difference, but I certainly do. I wish i didn't!

I have done abx mp3/wav testing before and spotted it every time.


oh i see. first you ripped your cds into format a, and now your ripping them again into format b. well that does make more sense, thank you for clearing that up. at first i thought you were "upsampling" lossy files into lossless, which is ridiculous.

 

i appreciate youre not trying to wind anyone up, and im not. neither am i trying to wind you up. but this is the science forum. if you come in here claiming to tell the difference between lossless and lossy - people are gonna want to see some proof, no one is going to take your word for it. infact, even if you did provide some abx results, they too could be questioned depending on the program you used for ripping the files and such, there are very meticulous rules to providing reliable test results.

 

you must understand, alot of people here, me included, have taken ABX tests and so far no one has succeeded in telling the difference. infact, if you follow the rules, and it turns out your ears are infact superior to ours, youll earn abit of fame around the internet. as far as i know, there have only been 1 or 2 people who have actually proved they can tell

post #18 of 76
Quote:
Originally Posted by KT66 View Post

i did not do abx testing, i did not do blind testing, and my post was not meant to wind anyone up.
I respect that we all hear differently, and some may not hear any difference, but I certainly do. I wish i didn't!

I have done abx mp3/wav testing before and spotted it every time.

 

The point about blind testing is that it removes a ton of cognitive biases not least being that when you know what you are listening to you can hear the legendary flaws in it. Nobody is immune to this. However removing the knowledge of what you are listening to frequently removes these obvious differences for even golden eared listeners with multi £K systems.

 

Perhaps your hearing is exceptional and you can reliably hear the differences between high quality lossy compressed files and lossless files frequently but to date not one human has ever been able to show a 100% detection rate for all samples. Some tracks are problematical for lossy encoders and the algorithms mangle parts giving rise to artifacts such as warbling or massive high frequency  roll-offs. That said for most people most of the time a high quality VBR 0 or CBR 320K Mp3 file is audibly indistinguishable from a lossless file in normal listening.

 

If you want to take over the mantle of resident golden ear currently held by UltMusicSnob you'll have to provide better proof

post #19 of 76
Quote:
Originally Posted by adamlr View Post
 

you must understand, alot of people here, me included, have taken ABX tests and so far no one has succeeded in telling the difference. infact, if you follow the rules, and it turns out your ears are infact superior to ours, youll earn abit of fame around the internet. as far as i know, there have only been 1 or 2 people who have actually proved they can tell

 

That is not true. There have been quite a few positive DBTs on some material (UMS here and his Ravel sample and others) and if you visit HydrogenAudio there are quite a few there as well. While rare it is by no means unheard of. Even my old ears have detected the difference between an Mp3 and a Wav file with a very hot  sample (Nobody Wants To: Crowded House)  - see below for wav and mp3 - red is clipping - note how the encoding adds a bunch of clipping

 

post #20 of 76
Quote:
Originally Posted by nick_charles View Post
 

 

That is not true. There have been quite a few positive DBTs on some material (UMS here and his Ravel sample and others) and if you visit HydrogenAudio there are quite a few there as well. While rare it is by no means unheard of. Even my old ears have detected the difference between an Mp3 and a Wav file with a very hot  sample (Nobody Wants To: Crowded House)  - see below for wav and mp3 - red is clipping - note how the encoding adds a bunch of clipping

 

first of all, thank you for correcting me. 2nd, ive heard the phrase "hot sample" once or twice before - what does it mean?

 

and last, im afraid i cant understand what you uploaded there (thats why i usually lurk here without posting :tongue_smile:) im assuming the clipping happened with the high frequencies? or is it something to do with squashed DR? i know about clipping from my experiments with equalization, where if you boost a frequency too much, it reaches (or passes) full scale, and clips. how would compression cause such a thing?

post #21 of 76
Quote:
Originally Posted by adamlr View Post
 

first of all, thank you for correcting me. 2nd, ive heard the phrase "hot sample" once or twice before - what does it mean?

 

and last, im afraid i cant understand what you uploaded there (thats why i usually lurk here without posting :tongue_smile:) im assuming the clipping happened with the high frequencies? or is it something to do with squashed DR? i know about clipping from my experiments with equalization, where if you boost a frequency too much, it reaches (or passes) full scale, and clips. how would compression cause such a thing?

 

A hot track It is a track where the recorded level is at or very close to full scale i.e 0db for a lot of the time i.e loud. For a CD 0db is maximum volume and -96db or so is the lowest possible level signal - I'm simplifying a bit but it will do for now - so on some tracks the signal is as loud as it possibly can be for a lot. My limited understanding is that It is not so much the fact of compression itself but the encoding algorithm that has problems with this as it approximates output levels at certain points based on the signal it receives (normally pretty well) but can actually increase signal levels marginally at some points and when the input signal is bang on max these increases can sometimes cause clipping 
 
Here is a better explanation from the legendary JJ (Jim Johnson) 

It's really very simple.

MP3 codecs are lossy. They remove information, and add noise. By definition, this changes the waveform. A waveform that is engineered to be within || of max/min before coding will probably exceed coding by some substantial amount.

Also, if you're that close in level measurement, you really are very likely to have intersample overs, too.

 

 

 

The image I posted shows the output levels of an uncompressed wav file (bottom) over time and for an mp3 file created from it(top) using the latest Lame Mp3 encoder - red means the signal is clipping. the original wav file does have a few bits of clipping but encoding it adds a lot more. The tool used to display the waveforms is audacity which is free and cool !


Edited by nick_charles - 11/6/13 at 4:24pm
post #22 of 76
Quote:
Originally Posted by KT66 View Post

Proton007, who mentioned detail? I assume your post is just taking the p#ss

Oooh. The anger.
The rest of your post is irrelevant, or are you saying that those who aren't native English speakers are better at science?

Welcome to the science section.
Edited by proton007 - 11/6/13 at 4:18pm
post #23 of 76

KT66 please post a track where it is, as you say, "day/night" between lossless and 320 kbps AAC.

 

 

Quote:
Originally Posted by KT66 View Post

I played one track. Then played the other, i rarely had to do it twice

Are you serious?

post #24 of 76
Quote:
Originally Posted by nick_charles View Post
 

The image I posted shows the output levels of an uncompressed wav file (bottom) over time and for an mp3 file created from it(top) using the latest Lame Mp3 encoder - red means the signal is clipping. the original wav file does have a few bits of clipping but encoding it adds a lot more. The tool used to display the waveforms is audacity which is free and cool !

 

You need to reduce gain when encoding to lossy formats to avoid clipping. Try encoding at -6dB - that should avoid it in all but the most extreme cases.

post #25 of 76
Quote:
Originally Posted by KT66 View Post

I played one track. Then played the other, i rarely had to do it twice

 

No offense but this method sounds like something that can be heavily biased with placebo mate.

post #26 of 76

I understand there are some people who can and can't differentiate between lossy and lossless (CD), but I am at a loss (pun intended) why it's so adamantly defended. Shouldn't we be pushing atleast a few levels closer to the original master whether that be tape, 24/96 or DSD and comparing lossy from there? :confused: 

 

It's like, "which of the lossy formats is lossi-er" in here. Just my 0.02


Edited by brunk - 11/6/13 at 6:40pm
post #27 of 76
Quote:
Originally Posted by brunk View Post
 

I understand there are some people who can and can't differentiate between lossy and lossless (CD), but I am at a loss (pun intended) why it's so adamantly defended.

This depends on the track, the codec, the encoder used and the quality setting. This is no defense but it seems like there's no such thing as transparency for audiophiles, despite it being well known in psychoacoustics.

 

 

Quote:
Shouldn't we be pushing atleast a few levels closer to the original master whether that be tape, 24/96 or DSD and comparing lossy from there? 

No, why not just use the CD?

 

 

Quote:
 :confused: 

 

It's like, "which of the lossy formats is lossi-er" in here. Just my 0.02

Well you can clearly rank codecs, encoders by transparency at a given bitrate. There are a couple of multi-format listening tests online.

post #28 of 76
Quote:
Originally Posted by xnor View Post
 

This depends on the track, the codec, the encoder used and the quality setting. This is no defense but it seems like there's no such thing as transparency for audiophiles, despite it being well known in psychoacoustics.

 

 

No, why not just use the CD?

 

 

Well you can clearly rank codecs, encoders by transparency at a given bitrate. There are a couple of multi-format listening tests online.

 

Well, CD is still a lossy format. For sake of sticking to thread topic, shouldn't we be comparing a true lossless master, from say Blue Coast or something to a mp3? I think the problem is that we are so entrenched in 16/44.1 and mp3 that we forget these are ALL mixdowns. Wouldn't you agree?

post #29 of 76
Quote:
 Originally Posted by xnor View Post
 

This depends on the track, the codec, the encoder used and the quality setting. This is no defense but it seems like there's no such thing as transparency for audiophiles, despite it being well known in psychoacoustics.

 

 

No, why not just use the CD?

 

 

Well you can clearly rank codecs, encoders by transparency at a given bitrate. There are a couple of multi-format listening tests online.

 

Quote:

Originally Posted by brunk View Post
 

 

Well, CD is still a lossy format. For sake of sticking to thread topic, shouldn't we be comparing a true lossless master, from say Blue Coast or something to a mp3? I think the problem is that we are so entrenched in 16/44.1 and mp3 that we forget these are ALL mixdowns. Wouldn't you agree?

I guess just to clarify, wouldn't it be more appropriate to get a true master from a reputable source like Blue Coast (Chesky's HDTracks is too spotty), then mix it down to a mp3 for a true and proper "lossless vs lossy" showdown. Let me know if I can somehow help collaborate/contribute with this, as you seem more capable than I in regards to "sound science".

:beerchug:

post #30 of 76
Quote:
Originally Posted by brunk View Post
 

 

Well, CD is still a lossy format. For sake of sticking to thread topic, shouldn't we be comparing a true lossless master, from say Blue Coast or something to a mp3? I think the problem is that we are so entrenched in 16/44.1 and mp3 that we forget these are ALL mixdowns. Wouldn't you agree?


No, maybe you mean something different but usually you always listen to a mixdown regardless of format.

 

44.1/16 is fine (covers the whole hearing range and has a high dynamic range) and is the most common format. Encoding something with a higher sample rate just introduces more variables into the comparison.

 

edit: What you could do is compare 96/24 to 44.1/16 and then compare 44.1/16 to lossy.


Edited by xnor - 11/6/13 at 7:40pm
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Sound Science
Head-Fi.org › Forums › Equipment Forums › Sound Science › Lossless vs. Lossy