Audibility of compressed vs uncompressed music

cnocbui

This is is a follow on from an issue raised in another thread, which I am continuiing here so as not to derail that thread:

http://www.boards.ie/vbulletin/showthread.php?p=76738889#post76738889

My assertion was that I have not come across anyone who could hear a difference between 223 kbs AAC compressed music and the uncompressed original.

Slaphead07 wrote: »

Really? Chalk and cheese on a real hifi system. Even 320kbs on a modest one.

Chalk and cheese eh?

Here is a WAV file (21mb):
https://rapidshare.com/files/3023221742/SourcePlus223AAC_short.rar

It consists of bits that are uncompressed and bits which are AAC compressed at 223 kbps.

If you - or anyone else who is interested - can tell me the time intervals where the edits are, I will acknowledge you can hear a difference. If not, well then obviously you can't.

Slaphead07

I really don't feel the need to prove anything to you. If you can't hear it then look to the equipment you're using or your own hearing.

Assert whatever you want but if you think you can permanently remove >75% of an audio signal and not effect the source then you're very much mistaken. How foolish of the music industry to spend so much time and money on producing CDs, not to mention high resolution files, without consulting you first.

cnocbui

My equipment is more than adequate, and I believe the same holds for my hearing, thanks.

You don't have to prove anything to me - unless you want me to credit you with the capabilities you claim.

My original assertion still holds. Still haven't encountered anyone who can hear the difference.

Slaphead07

so this assertion is all based solely on your expieriences?

Ok. Well I'm gonna go with the combined knowledge of the entire recording and hif industry on this.

cnocbui

My assertion is based on my experience of not personally being able to hear a difference and not having encountering anyone else who can either.

I am aware that a lot of other people have conducted double blind trials and can't hear a difference either. Public listening tests have been conducted via the hydrogenaudio.org forums for years (foobar) and the overwhelming conclusion of these has been that for the vast majority of listeners and music samples, no difference can be heard at even moderate compression levels.

A very few individuals have reported a repeatable/provable ability to hear a difference, but they are in a tiny minority. A few 'killer' samples have come to light where compression artifacts can be heard, but again, these are a tiny sub set of available material.

As for the Hi-Fi industry, Just about every manufacturer from B&W to krell offer iPod friendly devices. The recording industry: they have been willfully degrading their own product for decades in the well known 'loudness wars', which has lead to a vastly greater and distintctly audible reduction in quality than even lowish bit rate compression achieves, so their 'knowledge' doesn't seem relevant.

Personally, I go with what my ears report when deciding what I can and can't hear, not what hifi manufacturers or others claim I should be able to.

If someone claimed they could walk on water, I would like to see a demonstration before believing them. If someone stated they were capable of precognition and could win the Euro millions lottery whenever they wanted: again, I would like to see a winning ticket and the cheque before believing their claim.

Same goes for this topic. I won't believe anyone's assertion they can hear a difference unless they can clearly show they can. I am told the difference is like night and day and is painfully obvious, yet no one has ever been able to tell me where the edit points are in that file. Strange that.

dyer

most people really believe they CAN hear a difference between compressed vs original audio.. but then, these people haven't bothered really testing out that idea either because they just assume they'd be able to hear it.

just get a copy of foobar and the ABX plugin.. encode a few tracks with different bitrates youd like to check and away you go.

over V0 and in fact.. V2 mp3 id wager you probably wont score as high as you imagined you would, if at all.

Slaphead07

cnocbui wrote: »

I won't believe anyone's assertion they can hear a difference unless they can clearly show they can.

I can live with that. G'luck.

Fuzzy Clam

There's a change in audio quality at about 17secs anyway but without been able to compare it with the original uncompressed file I couldn't say for sure. Also possibly at 1.24.

ZENER

There are so many variables with this "test". Not least of which is the chipset and design of the audio system in the computer used.

What would even it up - and I'm prepared to be corrected here - would be to convert them and transfer them to an audio CD played through a good sound system and speakers. I believe that then a difference would be noticed - if any exists.

For myself I use an M-Box Firewire audio unit fed from a Powerbook which acts as a media server. The amplifier is a Sony AV amp and the Speakers are AE 100s. Playing AAC files through that I can definitely tell the difference between a compressed version of a song and the original CD. But as I said already, the quality and design of the MBOX or any sound card will colour the results.

I'll download that file, burn a CD and listen to it and report back.

Ken

cnocbui

Fuzzy Clam wrote: »

There's a change in audio quality at about 17secs anyway but without been able to compare it with the original uncompressed file I couldn't say for sure. Also possibly at 1.24.

You are correct for one edit, but not the other. There are however, several more, so I would have to say the correct hit could be random. I had a listen again to the file myself, concentrating on the 'change' you reported. I could 'see' what you mean, so I went to the original WAV and the same 'change' is there so I suspect it is an artifact of the mixing/editing. I can upload the original WAV segment for you if you would like.

Thank you for having a listen and having the courage to report on your findings.

andy1249

First , this is a good read

http://www.stereophile.com/features/308mp3cd/index.html

Technically at least , that should drive home the fact that regardless of the compression type , if the format is lossy , then in no way is it CD quality and never will be.

Second , can you tell the difference ?

FYI , before I begin my music listening setup is fairly modest , its a Marantz 8003 driving a pair of Monitor audio floorstanders , my file based playback is via Squeezbox touch and my CD playing equipment is an Oppo BD83.



In my experience , the only way an MP3 would give me difficulty in identifying it is with lame VBR 0 , I have been able to pick out any mp3 done at a fixed bit rate.

I know this because at one time , I had all 3000+ of my albums encoded to MP3.

After many many listens to my collection in lossy format , and feeling more and more depressed about the amount of time I had wasted , the decision was made to start the whole process over again , this time ripping to flac lossless and with a quality drive and software.
As you can imagine , this wasnt something I was about to do without being sure , even had some friends over to confirm what I already know , that lossy is just no damn good!


I never ripped an album at less than 320kbps, I re-ripped some at VBR 0 to see if it improved things , on some pop music content this made things better , but still not as good as the CD , and with classical content it was as clear as anything that what you were listening too was hugely compromised.

Now, some of them were not done very well at all , thats a given , the temptation is to rip them as quick as possible , encoding a 70 minute album in less than 4 minutes , this will always lead to multiple drive errors.

Software is also a major consideration , for example itunes at default settings is one of the absolute worst ways to rip a file , it always produces a file that is easily distinguishable from the CD, it does a very quick and very dirty job.

If your going to rip properly , you must use EAC or something similar , and this will take up to 50% or more of the actual playing time of the disc , even at that , differences can still be heard between a lossy encode format and the CD depending on the content.


A quality rip and lossless encoding produces results that are indistinguishable from a CD, and if you want to be certain that what your doing is going to be indistinguishable from the CD , for all music types , then this is the only way to go.

3DataModem

In terms of iTunes, it now gives you the ability to rip any old file and then have access to a high bit rate (320) file for stream or download. Very useful service called iTunes Match costing about 25 a year.

cnocbui

andy1249 wrote: »

First , this is a good read

http://www.stereophile.com/features/308mp3cd/index.html

Technically at least , that should drive home the fact that regardless of the compression type , if the format is lossy , then in no way is it CD quality and never will be.

A lossy file does not have the same 'quality' of data as an uncompressed file - that goes without saying.

Second , can you tell the difference ?

This of course is the important question.

In my experience , the only way an MP3 would give me difficulty in identifying it is with lame VBR 0 , I have been able to pick out any mp3 done at a fixed bit rate.

Can you identify the compressed sections of the file I uploaded though?

There is a very powerful psychological effect called the placebo effect. Basically this means if you 'know' something or 'believe' something, it can have an effect on your perception and it can even lead to a measurable physical effect on your body. That is why in drug trials they give some patients the real drug and some a placebo, because it is an established fact that just giving someone a pill that has no active drug in it and telling them it will have an effect, can and will lead to a significant effect in many people.

Taste testing at wine competitions are done so the tasters do not know what the wine they are tasting is, because if they did, it would heavily influence their judgment.

Scientists discovered another far more subtle and annoying problem. If the scientist knew who was getting the placebo and who the real drug, the subjects would in some cases pick up on the behaviour of the researcher and it would affect the results! In other words, some researchers would behave very subtly differently towards those on the placebo and those on the real drug and the test subjects would pick up on it and it would effect how they responded.

So scientists came up with a testing methodology called the 'double blind trial'. In a double blind trial, neither the person being tested nor the researchers administering the test, know which pill is a placebo or which is real. Only after the trial is over do the researchers get to find out who got what.

So when a person says they have no problem telling a compressed track from uncompressed, when their listening is done in a way they know what they are listening to, their statement is meaningless from a scientific point of view.

You get people who say the difference is like night and day, even for 320kbps, yet when asked to tell the difference when they don't know what they are listening to, suddenly their ability to tell night from day, chalk from cheese - vanishes.

There is a software package called Foobar, as dyer pointed out, that has an ABX module. Basically you rip a lossless track and it makes a lossy version at the bit rate you are testing, then the software plays back bits of the lossy track and bits of the uncompressed, in a random order, sometimes playing each type twice, and the listener has to say which is which. It does this multiple times and then delivers a statistical report which indicates whether the person could distinguish between the two better than random chance would allow.

Basically Foobar runs a double blind trial.

The results from a lot of people testing their ability to distinguish between compressed and uncompressed using Foobar, is overwhelmingly that most people can not perceive a difference at even moderate bit rates, let alone high bit rates. If you don't believe me, have a look at the relevant threads where these public listening tests have been documented and their results reported on:

http://www.hydrogenaudio.org/forums/index.php?showforum=40

Now I am not saying you definitely can't hear a difference or that it is impossible that you can, because a very few people have shown statistical results from foobar that indicate they clearly can. You could be one of them.

I know this because at one time , I had all 3000+ of my albums encoded to MP3.

After many many listens to my collection in lossy format , and feeling more and more depressed about the amount of time I had wasted , the decision was made to start the whole process over again , this time ripping to flac lossless and with a quality drive and software.
As you can imagine , this wasnt something I was about to do without being sure , even had some friends over to confirm what I already know , that lossy is just no damn good!

When I came to the conclusion I couldn't hear a difference, I thought it was just me and that my hearing was not good enough, so I had a few other people listen the same way I did and none of them could hear a difference either. It wasn't a double blind trial, I knew what they were listening to, but they didn't.

Basically I had my iPod and CD player simultaneously playing the same track so they were synced time-wise. My pre-amp allows you to match the volume levels of inputs (which is very important) and then I just switched between the sources while the track was playing and asked if one sounded better than the other.

This is the system we listened on:

Software is also a major consideration , for example itunes at default settings is one of the absolute worst ways to rip a file , it always produces a file that is easily distinguishable from the CD, it does a very quick and very dirty job.

iTunes is actually a very good way to rip music and it's codecs are very good. You can nominate what bitrates to use and whether VBR or not.

A quality rip and lossless encoding produces results that are indistinguishable from a CD, and if you want to be certain that what your doing is going to be indistinguishable from the CD , for all music types , then this is the only way to go.

The magazine 'The Absolute Sound" apparently published a recent article in which they stated Flac wasn't as good as CD :rolleyes:

They also claimed that if you took a WAV file, converted it to Flac and then converted it back to WAV, there was a definite loss of quality. :rolleyes:

Now I am not saying people should listen to lossy rips if they don't want to - HD storage is pretty cheap these days so there is no reason not to rip losslessly. It is probably quicker. Storage on mobile devices like phones and PMPs is somewhat less cheap, so a lot of people might want to use compression to fit more on them. I know I do.

Fuzzy Clam

cnocbui wrote: »

You are correct for one edit, but not the other. There are however, several more, so I would have to say the correct hit could be random. I had a listen again to the file myself, concentrating on the 'change' you reported. I could 'see' what you mean, so I went to the original WAV and the same 'change' is there so I suspect it is an artifact of the mixing/editing. I can upload the original WAV segment for you if you would like.

Thank you for having a listen and having the courage to report on your findings.

thanks cnocbui.
I'm confident that the audio after 1.24 is compressed.I don't think that one was a random guess so I'll claim my prize:D. I was unsure about the 17secs.

Having said that, the overall quality was very similar. Slightly less top end on the vocal but none of the artifacs that we attribute to poor compression.

mcwhirter

cnocbui wrote: »

A lossy file does not have the same 'quality' of data as an uncompressed file - that goes without saying.

This of course is the important question.

Can you identify the compressed sections of the file I uploaded though?

There is a very powerful psychological effect called the placebo effect. Basically this means if you 'know' something or 'believe' something, it can have an effect on your perception and it can even lead to a measurable physical effect on your body. That is why in drug trials they give some patients the real drug and some a placebo, because it is an established fact that just giving someone a pill that has no active drug in it and telling them it will have an effect, can and will lead to a significant effect in many people.

Taste testing at wine competitions are done so the tasters do not know what the wine they are tasting is, because if they did, it would heavily influence their judgment.

Scientists discovered another far more subtle and annoying problem. If the scientist knew who was getting the placebo and who the real drug, the subjects would in some cases pick up on the behaviour of the researcher and it would affect the results! In other words, some researchers would behave very subtly differently towards those on the placebo and those on the real drug and the test subjects would pick up on it and it would effect how they responded.

So scientists came up with a testing methodology called the 'double blind trial'. In a double blind trial, neither the person being tested nor the researchers administering the test, know which pill is a placebo or which is real. Only after the trial is over do the researchers get to find out who got what.

So when a person says they have no problem telling a compressed track from uncompressed, when their listening is done in a way they know what they are listening to, their statement is meaningless from a scientific point of view.

You get people who say the difference is like night and day, even for 320kbps, yet when asked to tell the difference when they don't know what they are listening to, suddenly their ability to tell night from day, chalk from cheese - vanishes.

There is a software package called Foobar, as dyer pointed out, that has an ABX module. Basically you rip a lossless track and it makes a lossy version at the bit rate you are testing, then the software plays back bits of the lossy track and bits of the uncompressed, in a random order, sometimes playing each type twice, and the listener has to say which is which. It does this multiple times and then delivers a statistical report which indicates whether the person could distinguish between the two better than random chance would allow.

Basically Foobar runs a double blind trial.

The results from a lot of people testing their ability to distinguish between compressed and uncompressed using Foobar, is overwhelmingly that most people can not perceive a difference at even moderate bit rates, let alone high bit rates. If you don't believe me, have a look at the relevant threads where these public listening tests have been documented and their results reported on:

http://www.hydrogenaudio.org/forums/index.php?showforum=40

Now I am not saying you definitely can't hear a difference or that it is impossible that you can, because a very few people have shown statistical results from foobar that indicate they clearly can. You could be one of them.

When I came to the conclusion I couldn't hear a difference, I thought it was just me and that my hearing was not good enough, so I had a few other people listen the same way I did and none of them could hear a difference either. It wasn't a double blind trial, I knew what they were listening to, but they didn't.

Basically I had my iPod and CD player simultaneously playing the same track so they were synced time-wise. My pre-amp allows you to match the volume levels of inputs (which is very important) and then I just switched between the sources while the track was playing and asked if one sounded better than the other.

This is the system we listened on:




iTunes is actually a very good way to rip music and it's codecs are very good. You can nominate what bitrates to use and whether VBR or not.

The magazine 'The Absolute Sound" apparently published a recent article in which they stated Flac wasn't as good as CD :rolleyes:

They also claimed that if you took a WAV file, converted it to Flac and then converted it back to WAV, there was a definite loss of quality. :rolleyes:

Now I am not saying people should listen to lossy rips if they don't want to - HD storage is pretty cheap these days so there is no reason not to rip losslessly. It is probably quicker. Storage on mobile devices like phones and PMPs is somewhat less cheap, so a lot of people might want to use compression to fit more on them. I know I do.

This is an interesting discussion for me anyhow, as I have the same cd player as you, its a stage 6 (with changed caps, op amps etc). Now its a really great cd player but after I went back to vinyl it's hard to go back to cd(haven't tried newer cd players yet, so it could be player has had its day). For instance I have an album on both cd and vinyl, the vinyl version definately sounds more natural and has more realism. But then again, this could be to do with the mastering on both vinyl and cd.
But as for compression, it should be noticeable on a good system(ie, not via pc speakers). I don't enjoy listening to my ipod through the hifi, just too much of the audio is missing for me. But then, it could be my interface that is letting the side down.
And another point is as we get older , our hearing range diminishes so the people who can most probably hear the difference between your recordings will possibly not be able to tell you as they can't talk yet.

Like the speakers BTW

cnocbui

I suspected my hearing might be at fault, which is why I got my children to listen to my original iPod vs CD evaluation. Using a tone generator through the hifi, I confirmed they could both hear to 20khz. My hearing only went to 16khz.

I also had my son listen to the test file, but he couldn't detect the edits.

cnocbui

In this thread: http://www.boards.ie/vbulletin/showthread.php?threadid=2056554893


Slaphead said this:

Slaphead07 wrote: »

Uh huh? In a previous "discussion" you stated nobody could differentiate between a CD and a compressed audio file. When you posted a pic of your system I understood why. You stuck expensive speakers on the end of a mediocre and tired combination of amps and an Ipod as a source. Your system is simply incapable of differentiating between a red book CD recording and a file that's a fraction of the original quality. THAT has been your experience. There are no speakers that can take a poor source (i.e. compressed files) and restore them to full CD quality. They don't exist. Please accept that your experience is just that - it's not fact.

I don't mind you convincing yourself that your system is the bees knees but when you advise others to follow that path I have an issue with it.

Mediocre amps?

The pre amp is a Meridian 201. I don't believe Meridian make mediocre equipment. I doubt very many people would share your opinion that they do.

The Poweramp is a Perreaux 6000B with Balanced line inputs and XLR connectors.

Specifications:
Power Output
300Wrms per channel continuous both channels driven into 8Ω from 20Hz to 20kHz at no more
than 0.09% Total Harmonic Distortion.
500Wrms per channel continuous into 4Ω
Total Harmonic Distortion....................................................................................................< 0.09%
Intermodulation Distortion...................................................................................................< 0.09%
Bandwidth................................................................................................±0.25dB from 10Hz-20kHz
........................................................................................................................-3.0dB 10Hz and 50kHz

These amps are professional level equipment and are still used in recording studios for monitoring. They are also renowned for their longevity and build quality.

NRG recording studios - under list of equipment:

MONITORS
Genelec 1031A (2)
Tannoy SGM 10 w/Mastering Lab Crossovers (1)
Dynaudio BM15 (6)
Dynaudio BM15 Air Series (1)
Dynaudio BM 6 (4)
Auratone (1)
JBL Power 15 Eon PA Monitors (4)
Bryston Power Amp (1)
Dynaudio BM 6A (2)
Perreaux 3000B PWR Amp (1)
Perreaux 6000B PWR Amp (1)

http://www.nrgrecording.com/studios.php

The New Zealand school of music:

Lilburn Electroacoustic Music Studios

Electroacoustic Music Studio 1

The primary composition studio, a fully soundproof, air conditioned environment with superb monitoring, mixing and processing facilities. The studio has been substantially upgraded this year, offering state-of-the-art facilities for electroacoustic composition.
Mixing and monitoring

• Pro Tools HD-2 Accel hardware/Pro Tools 7 software
• Focusrite Control 24
• Digidesign 192 I/O digital interface
• Digidesign 888/24 digital interface
• Genelec 5.1 Surround Monitoring with Sub
• Tannoy FSM loudspeakers/Perreaux 6000B power amplifier
• 8-channel studio expansion, with additional Tannoy CPA.15 monitors

Here is a thread discussing them, including some nice pics,
http://www.audioheritage.org/vbulletin/showthread.php?16823-Perreaux-6000B&highlight=

I used to work in a professional video and audio production facility in Australia. When I was buying my amp SH, I got the techs to give it the once over. They said it measured better than the Crown amps our studio was using.

The speakers are B&W 802 Nautilus.

I also have a Micromega Stage 2 CD player. That is the source I used for my original comparisons with the iPod.

No doubt you consider that to be substandard too:

I recognised this as a player capable of well above average sound quality...

Measurements were pretty much state-of-the-art, with signal-tonoise ratio up around 96dB, channel separation about the same figure right across the audio spectrum, frequency response dead flat and low-level linearity maintained to within 1dB down to -90dB....

To sum up, this new integrated player from France in its Stage 2 version has the advantages of a compact, nicely styled single-box design, innovative thinking in relation to operational features and a technical performance in the highest class.

http://www.gramophone.net/Issue/Page/July%201994/128/810101#header-logo

As for the iPod. A complete Stereophile review of my iPod by John Atkinson in Stereophile:http://www.stereophile.com/budgetcomponents/934/index.html

The iPod's measured behavior is better than many CD players—ironic, considering that most of the time it will be used to play MP3 and AAC files, which will not immediately benefit from such good performance. But if you're willing to trade off maximum playing time against the ability to play uncompressed AIFF or WAV files, the iPod will do an excellent job of decoding them. Excellent, cost-effective audio engineering from an unexpected source.—John Atkinson

Your opinion that my system is incapable of differentiating between compressed and uncompressed music appears to have been formed from fantasies. Objective real world criteria would suggest the opposite was true.

The file I posted can be played on a computer, using HQ headphones for listening - which I also have, and which I have done - so where that is concerned, the quality of my 'mediocre' equipment is not relevant - though the main system is better.

Since you obviously think you have better equipment and hearing, why are you unable to hear the difference between compressed and uncompressed? Due to the nature of your assertions, I think saying you 'don't have to prove anything' is no longer the case.

Put your ears and your equipment where your mouth is.

Slaphead07

These amps are professional level equipment and are still used in recording studios for monitoring.

that sums it up for me. None of the qualities looked for in a high end hifi system, just detail and volume.

The pre amp is a Meridian 201. I don't believe Meridian make mediocre equipment.

Oh God they do. Although they've made some horribly harsh equipment too.

Since you obviously think you have better equipment and hearing, why are you unable to hear the difference between compressed and uncompressed? Due to the nature of your assertions, I think saying you 'don't have to prove anything' is no longer the case.

this makes no sense to me. I never said I had better hearing than you but I can tell the difference between CD and compressed the same as I can tell the difference between a black and white movie and a colour one or night and day. It's bloody obvious. I feel no pressure to prove anything to you. Ever.

viperirl

Slaphead07 wrote: »

Oh God they do. Although they've made some horribly harsh equipment too.

I wouldnt agree with that. If anything, Meridian are noted for having a laid back sound. Maybe the older stuff was harsh but I never heard it. Their CD players are well regarded but I wouldnt be gone on their active speakers.

Slaphead07

viperirl wrote: »

I wouldnt agree with that. If anything, Meridian are noted for having a laid back sound.

Granted... but this is older stuff we're talking about and a pre-amp in particular.

BTW, this was never meant to be one system Vs another, my point is that it's very easy to hear the difference between compressed files and red-book CDs. Unless one's system is unable to reveal that.

Fuzzy Clam

cnocbui wrote: »

Since you obviously think you have better equipment and hearing, why are you unable to hear the difference between compressed and uncompressed? Due to the nature of your assertions, I think saying you 'don't have to prove anything' is no longer the case.

Put your ears and your equipment where your mouth is.

For the record, when I listened to the file (and heard the difference in one instance) , it was using the bog standard on board sound card fed into cheap Thrust PC speakers and a pair of 21 Euro Sony MDR-V150 headphones!