Spectrograms before and after conversion to flac

[Cornucopia]

I'm with Slipster - Sounds like I'm going to have to pull out the old "CD vs. MP3 ABX test" file also so these "golden ears posers" can get off their high horse.

Listen to the 6 wav files and rate them, without looking at spectrograms or wavefrom monitors. Some are original WAVs, some high-bitrate compressed, some low-bitrate compressed, some losslessly compressed, some randome duplicates. Won't tell you more beyond that.
Music provided has little very-high-freq elements, a minus, but does have good acoustics with telltale reverb tails, a plus. So should be a decent source for comparison. (My own recording)

The first time I created these was in response to someone claiming the same thing about transparent compression being supposed garbage. This was a very definitive argument against that assertion.

Sure, anyone can not use the correct steps and create junk by accident, even when it shouldn't be junk. But those are exceptions to the rule.

It is a reproducable fact: Very High bitrate lossy compression is audibly/visibly transparent to the GREAT MAJORITY of general audiences. Lower bitrate lossy compression is NOTICEABLY NOT TRANSPARENT, and there's a range in-between where it is transparent to some and not to others.

Scott

edit: BTW, the original spectrogram business probably had to do with the fact that not only was Audacity dithering, but it was probably NOISE-SHAPING, which puts a higher percentage of the dither in the upper frequency range where human hearing is least sensitive.

[aedipuss]

i don't have any real problem with mp3 as a third generation audio format. it's their source material that is the problem. cds have audio holes wide enough to drive trucks through. people have heard nothing better, choose not to hear them, are anesthetized to them or have damaged hearing. pcm wav is a misnomer, there is no continuous wave, it's discrete points separated by empty space.

[Cornucopia]

aedipuss, you may be the audiophiles' audiophile, but what you're saying about CDs & LPCM wav is nonsense.

It's a law of physics (sampling theory) that #1, any bandwidth may be faithfully (aka 100% accurately) sampled if the sample rate is at least twice the bandwidth, and #2, any error in the level of sampling diminishes with the bit depth (# of bits used to represent a quantization level) with 16bit surpassing the best that LP & cassette & FM have to offer and 20+ bits surpassing the best that 2" master reels (with Dolby S-NR) have to offer. With HIGH-QUALITY A-to-D and D-to-A, digital faithfully & correctly duplicates/parallels an all-analog R/P chain.

Good example: I was honored to be witness to a recording testing session in 2005 where a top flight music department at a major university was trying to decide what new system to purchase for it's recording department. 20+ famous recording engineer/producers (all self-proclaimed "golden ears", above repute, and self-admittedly better than me) were invited to listen to ~6 of the best systems available and compare and rate these systems in order to arrive at the best system possible (money wasn't really a limiting factor here). These systems included both analog and hi-rez digital systems, and the comparison included both live classical & jazz musicians and time-delayed (both analog and digital) and recorded-playback of these musicians. This was done over a whole day (with breaks in between so that peoples' ears would have a chance to rest). While they did attempt to rank them, there was no real clear winner - even the live analog wasn't in any way measurably better. In the end 2 digital systems were recommended and one purchased and is in current heavy use (I wanted the Digidesign ProToolsHD to be chosen, but it was the 2nd one and afterwards was NOT purchased). Like the old "is it live or is it memorex", if they can't tell the difference, I don't see how you could claim that. Even the wonderfully nice sound system you mention is dwarfed by the one(s) in use by these guys. I'm talking systems that cost upwards of $50,000 (with speakers alone being $5-10k apiece).

Maybe you don't quite understand it, but "discreet points" in digital land are NOT separated by empty space. They are like contiguous stairs: INFINITESIMALLY SMALL stepped stairs. So small in difference to each other that it looks and feels (and SOUNDS) like a smoothly rolling plane (aka WAVEFORM).

Without some example (that is not qualified by having bad mastering/engineering errors - of which there are plenty of occurrences both analog & digital), you really shouldn't be talking about "holes big enough to drive a truck through". Talk about exaggeration.

Scott

[Mephesto]

PS Just for fun, I've uploaded two 44.1kHz 16-bit WAV files HERE. They're 'zipped', so you may need to install WinRAR to extract them. They may or may not be the same, so some of you may or may not hear a difference between them. I'd be interested to know whether any of you think you can and, if so, in what way you think they sound different.

I ABX'd this and got 25 out of 37 for Test-1.wav which I believe is the compressed MP3 at either 256 VBR or 320 CBR. The difference is extremely difficult to discern but it is there. My ******* head hurts now.

So am I right?

[aedipuss]

Without some example (that is not qualified by having bad mastering/engineering errors - of which there are plenty of occurrences both analog & digital), you really shouldn't be talking about "holes big enough to drive a truck through". Talk about exaggeration.

hehe yes they would have to be microbot trucks. but the analogy holds. digital music is stair-stepped square pattern that approximates the analog original. just not closely enough for some. i sold high end audio gear for years and have worked in an audio recording studio. it's just my preference to hold out hope for a "closer" approximation someday than 44.1 16 bit.

here's my truck driving through a digital audio hole. could just as likely be an strat violin that lost it's unique timbre in there.

[aedipuss]

HERE[/URL]. They're 'zipped', so you may need to install WinRAR to extract them. They may or may not be the same, so some of you may or may not hear a difference between them. I'd be interested to know whether any of you think you can and, if so, in what way you think they sound different.

test-1.wav sounds a bit muddier and has lost some high frequency brilliance compared to test-a.wav

[Slipster]

Well spotted, those who listened!

That was the one CD in my collection of 200+ where I can hear any discernible difference without the need for an ABX test when using LAME at -V2 (~190kbps), so I'd kind of hoped that any other keen-eared listeners would notice it too. Encoding that particular album at -V0 fixes it for me but may not for others. Consider this particular sample pair as my personal method of sorting the "It sounds crap because I say so!" brigade from those who care enough to actually test this stuff out.

Check out Cornucopia's sample set for further testing if you're so inclined as he's probably a better provider of typical samples than I'm going to be as my musical tastes are anything but typical.

With regards the stepped waveform of digitally stored music, that never happens on playback due to the DAC performing oversampling that makes the steps virtually disappear. If you hook up an oscilloscope to a CD player or PC soundcard output and see anything even vaguely resembling the image above, the hardware is very badly broken.

You're unlikely to find a DAC that doesn't at least 4-times oversample these days, and a 4-times oversample halves, then halves, then halves, then halves again the step distance in both the X and Y directions meaning that any errors introduced by the steps themselves fall well outside the human ear's resolution in terms of amplitude step size and frequency content. 256-times oversampling is fairly common now in high-end DACs.


In short, with a sample rate of 44.1kHz, put a 20kHz sinewave in and you'll get a 20kHz sinewave out as close as makes no difference to a human being. Nothing more is required as you can't hear anything above 20kHz. Check out the Wiki entries for 'Nyquist-Shannon sampling theorem' and 'Oversampling' for further info. Both Wiki entries are correct and show the maths behind the arguments.

[pandy]

here's my truck driving through a digital audio hole. could just as likely be an strat violin that lost it's unique timbre in there.

You forget (intentionally or not) RECONSTRUCTION FILTER which is MANDATORY from sampling point of view - remember that correct analog to digital then digital to analog system imply at the input correct ANTIALIASING filter and correct RECONSTRUCTION filter at the output.

And there is no big trucks and big holes - this is pseudoscientific blabling .

Yes there are discrete values in discrete times but until whole system is correctly designed You can reconstruct ANY complex (magnitude and phase) signal with accuracy of single time/level quantization step.
With perfect antialiasing and reconstruction filters You need only two samples with less than perfect You need more than 2 samples.

[Ankin]

Hey Everyone.
I'm gonna bump this question of mine. It kinda got lost in the heat of the whole vinyl vs CD vs Mp3 CBR vs Mp3 VBR debate

Now that I'm aware of dithering, it kinda poses another question. Wikipedia says that dithering is used when digital audio is reduced to 16 bits for pressing onto a CD..
When I produce a flac file in Audacity, the output bit depth of the created file is 16 bits. Now that dithering is off, does this mean that I should go to Edit -> Preferences -> Quality -> Sampling and change the Default Sampling Format from 32-bit float to 16 bits as well?

Would really appreciate your advice..

[pandy]

Hey Everyone.
I'm gonna bump this question of mine. It kinda got lost in the heat of the whole vinyl vs CD vs Mp3 CBR vs Mp3 VBR debate

Sorry for that.

Now that I'm aware of dithering, it kinda poses another question. Wikipedia says that dithering is used when digital audio is reduced to 16 bits for pressing onto a CD..
When I produce a flac file in Audacity, the output bit depth of the created file is 16 bits. Now that dithering is off, does this mean that I should go to Edit -> Preferences -> Quality -> Sampling and change the Default Sampling Format from 32-bit float to 16 bits as well?

Would really appreciate your advice..

Seems that this is unavoidable - looks like Audacity convert all files to "default" format thus when You saving it will imply requantization and this is why dithering seems to be used.
So short answer - Yes, You need to change all Audacity settings to match Your source files.
It is strange that Audacity not provide to user chance to avoid conversion - it should be some switch to turn off all such features.

However may i ask why You use Audacity instead directly converting Your source to FLAC?

[Cornucopia]

Assuming this is what might have happened:
If Audacity takes a 16bit wav file and opens it with a default setting of 32bit, it's now padded those extra bits with ZEROs. But if it does ANY processing at all - even very minor, upon saving/exporting it will need to get the bit depth BACK DOWN to 16bit (unless 32bit is specified as the output). This is where dithering would make sense (as the alternative is rounding or just plain truncating = bad). And as I mentioned before, if the type of dithering includes NOISE SHAPING, that would account for the added High-Frequency energy.

Scott

[Ankin]

However may i ask why You use Audacity instead directly converting Your source to FLAC?

No special reason, I'm using it simply because I'm used to it.

If Audacity takes a 16bit wav file and opens it with a default setting of 32bit, it's now padded those extra bits with ZEROs. But if it does ANY processing at all - even very minor, upon saving/exporting it will need to get the bit depth BACK DOWN to 16bit

Ok, but now that I've changed the default settings to 16 bits? Will it still change the source to 32 in order to do some processing and then reduce it to 16 bits, or will it do everything in a '16-bit mode'?

Because basically, all I need is this:
I have a couple .aac and .ape files. I want to cut some unnecessary stuff out of them (at the beginning and at the end) and re-save them as .flac without introducing any change to the part that I want. Is there any way I could still use Audacity for that, or should I switch to something else?

[pandy]

Assuming this is what might have happened:
If Audacity takes a 16bit wav file and opens it with a default setting of 32bit, it's now padded those extra bits with ZEROs. But if it does ANY processing at all - even very minor, upon saving/exporting it will need to get the bit depth BACK DOWN to 16bit (unless 32bit is specified as the output). This is where dithering would make sense (as the alternative is rounding or just plain truncating = bad). And as I mentioned before, if the type of dithering includes NOISE SHAPING, that would account for the added High-Frequency energy.

Scott

Are You sure? - Audacity native (default) is 32 bit float - i'm not sure that conversion between 16int<>32 float is fully transparent...

[FulciLives]

Any number of tools can convert AAC and APE to FLAC but if you want to edit (cut the start or end bits out) then you will need to use an audio editor such as Audacity or GoldWave etc.

[showtaper]

Well spotted, those who listened!

That was the one CD in my collection of 200+ where I can hear any discernible difference without the need for an ABX test when using LAME at -V2 (~190kbps), so I'd kind of hoped that any other keen-eared listeners would notice it too. Encoding that particular album at -V0 fixes it for me but may not for others. Consider this particular sample pair as my personal method of sorting the "It sounds crap because I say so!" brigade from those who care enough to actually test this stuff out.

Check out Cornucopia's sample set for further testing if you're so inclined as he's probably a better provider of typical samples than I'm going to be as my musical tastes are anything but typical.

With regards the stepped waveform of digitally stored music, that never happens on playback due to the DAC performing oversampling that makes the steps virtually disappear. If you hook up an oscilloscope to a CD player or PC soundcard output and see anything even vaguely resembling the image above, the hardware is very badly broken.

You're unlikely to find a DAC that doesn't at least 4-times oversample these days, and a 4-times oversample halves, then halves, then halves, then halves again the step distance in both the X and Y directions meaning that any errors introduced by the steps themselves fall well outside the human ear's resolution in terms of amplitude step size and frequency content. 256-times oversampling is fairly common now in high-end DACs.


In short, with a sample rate of 44.1kHz, put a 20kHz sinewave in and you'll get a 20kHz sinewave out as close as makes no difference to a human being. Nothing more is required as you can't hear anything above 20kHz. Check out the Wiki entries for 'Nyquist-Shannon sampling theorem' and 'Oversampling' for further info. Both Wiki entries are correct and show the maths behind the arguments.

Except for the affect that harmonics have against the original. Also the very steep filters used to keep the frequencies within the sample window cause phase shifting in the signal which can be audible. Unless "noise" (dither) is added, there are audible artifacts down at the low end (level, not frequency). Higher resolution sampling rates reduce (but not eliminate) these phenomena.

Digital in any version WAV, mp3, etc, is quite easy to recognize if one knows what to listen for. I do agree that the younger generation has been trained to accept this as "perfect" audio reproduction, which IT IS NOT. Analog has its flaws as well, but it will always be my preferred format for serious listening.

BTW - think of wikipedia articles as a starting point for serious reading about anything. They are the "Windows for Dummies" equivalent of actual research.

BTW2 - Who do you think stacks the deck on all those mp3 shootouts? mp3 is a nice portable cannister - it was never meant to replace higher resolution formats or the pros would be recording with ipods.

[Ankin]

I do agree that the younger generation has been trained to accept this as "perfect" audio reproduction, which IT IS NOT.

I was under the impression that the older generation has gotten used to hearing vinyl artifacts..

Digital in any version WAV, mp3, etc, is quite easy to recognize if one knows what to listen for.

So you're actually making a claim that you can easily recognize a WAV copy of a vinyl recording?

[pandy]

Except for the affect that harmonics have against the original. Also the very steep filters used to keep the frequencies within the sample window cause phase shifting in the signal which can be audible. Unless "noise" (dither) is added, there are audible artifacts down at the low end (level, not frequency). Higher resolution sampling rates reduce (but not eliminate) these phenomena.

Once again - seems that there is a lot of misunderstanding of idea how digital system works.

- harmonics - they are everywhere - analog and in digital world (and if they are not correlated with signal then they should be not a problem for even golden ear guy (what with harmonics or nonlinear distortions produced in almost any physical device that convert electrical current to variable air pressure i.e. speakers, headphones etc),

- very step filters are only necessary when Nyquist converter is used (and seems that there is lot of golden ears guy that believe that TDA1543 with Nyquist sampling is holly grail for audio reproduction - no oversampling at all! so called NOS) - for normal case (multiple oversampling) this is no longer true, also digital filters can be designed to have linear phase response (and most cases they are linear phase but if You prefer analog sound You can use digital approximation of the Bessel filter which is almost never used by analog guys anyway - too complicated, not efficient enough - some of You guys are quite happy with Chebyshev filters and most of You use Butterworth filters),

- noise (dither) is in analog signal too (bias HF in magnetic recording is special case for dither, btw all capacitors store energy in a discrete way - or perhaps i should say electric charge have particular granularity) - it will linearize quantizer and spread quantization errors over whole spectrum (or with noise shaping pushed nonlinearities outside audible area where they are filtered and turned to heat).

Digital in any version WAV, mp3, etc, is quite easy to recognize if one knows what to listen for.

Usually yes but only due (in many cases) its transparency (so called lifeless, clinically cold sound) over typical consumer analog quality (nice warm, fat, bright sound) .

Analog has its flaws as well, but it will always be my preferred format for serious listening.

I would say that not many sources for audio (today) are analog (some small, niche recording companies?) thus hearing digital audio is almost unavoidable.

BTW - think of wikipedia articles as a starting point for serious reading about anything. They are the "Windows for Dummies" equivalent of actual research.

Usually Wikipedia articles are based on research results which are usually provided as a source so - Wikipedia is a quite good starting point for searching for a knowledge. Most of claims about trucks, steps, analog linearity etc looks like "Golden ears hearing for Dummies" - no offense but there is lot of hype behind all this golden-ear fashion - i know it is good to be superhearinghuman but in most cases this is not true.

I will say this once again - if golden ears guys can deal with few % distortions in speakers, headphones, silver wired audio transformers for tube amplifiers (not many vacuum tube amplifiers use OTL configuration) then they can live also with 0.00002% digital distortions related for quantization errors (i assume jitter free, 16 bit converter).

[Slipster]

Who do you think stacks the deck on all those mp3 shootouts?

In the case of the independent tests, nobody. There have been several of those over the years carried out by competent online groups who do everything possible to make sure that no bias comes into the testing and result presentation.

Tests still occur as one-offs on an almost weekly basis on another forum I visit occasionally when people turn up saying "MP3 is crap!" only to be asked to take an ABX test and, upon taking a test, fail it quite convincingly. The sad fact is that so many turn up with the "MP3 is crap!" attitude and refuse to take a test, thus carrying on to spread the message with no practical evidence on their part to support their claim.

mp3 is a nice portable cannister - it was never meant to replace higher resolution formats or the pros would be recording with ipods.

No they wouldn't, and that's a totally different argument. Lossy encoding has never made the claim of being lossless. That's why it's called lossy. It would be totally ridiculous to use any format for studio purposes that couldn't take multiple 'bumps' with as close to zero loss of quality as possible, hence the use of lossless formats.

The final delivery format only needs to be as good as the weakest link in the chain. That's why there isn't a single accurately documented case of anyone being able to ABX between 44.1/16 and anything above it as, regardless how good someone might think their ears and equipment are, properly encoded and decoded 44.1/16 exceeds the limitations of the human ear.

The same won't apply to any lossy codec all of the time under all circumstances, but using an encoder as intended sure helps! It's just the eejits who say "Lossy is broken! Period!" who wind me up, especially if they're totally ignoring advice from the codec developers regarding which settings to use and which to avoid.

I am aware of and do understand the points you bring up above re phase errors near cut-off points, etc, but the fact of the matter is that a listening test is a listening test, and no amount of pontificating alters the result of what an individual hears unless it helps them to train themselves to spot the weaknesses of a lossy system.

Some listeners are more susceptible than others to picking up these weaknesses with little or no training, but that still doesn't alter the fact that hardly anyone can tell a well encoded lossy file from a lossless one even in a double-blind listening test. Either you trust the results of ABX tests or you don't, but I trust them on the basis that they tell me whether I can actually hear a difference or not rather than theorising that I should be able to when, nearly always, I can't.

[Cornucopia]

@Slipster, I'll have to take issue with you re:ABX vs. 16/44. Ear/Brain doesn't just work in Amplitude+Frequency domains but also in Time domain. In this respect, 24/96 or 24/192 IS clearly & noticeably better, and I can tell the difference with phase information (placement) in Stereo/Surround programs and in the reverb tails. Once you're trained to work with those particular variables, the difference isn't minute. Otherwise, what you say is quite valid.

Me, I'd still use 24/96 or 24/192 to master with - that way degradation can occur but never build up to the point of messing up transparency. For most end uses, 16/44 or 16/48, or maybe 20/48 (with the correct high quality chain) are plenty good.

More later,
Scott

[turk690]

What percentage of typical consumer amps are Class D? I'd gladly take a guess at less than 1% with the other 99% being Class B or Class A/B. If you're talking about crossover distortion in simpler Class B designs then it may be audible in a very poorly designed amplifier, but designs that bad are very few and far between nowadays. 30 years ago, maybe. Now? Only if the amp fell out of a Christmas cracker.

It may well be higher than 1%. For example, I'm using now a Logitech Z623 system on this PC which can put out 200Wrms and is THX-certified. Class D amplifiers are more efficient (less heatsinks mean smaller dimensions) and can be found in computer audio systems, car audio, all-in-one entertainment centers, probably even the audio amplifier stage of your smart phone.

[Slipster]

It may well be higher than 1%...

You're right of course, not that being Class D is actually a problem in terms of switching noise reaching the output if you look at the spec sheets for most of the recent integrated designs. I guess the point I was trying to make was that no amplifier design has transistors in the signal path that switch on and off unless it's inherent to their design, ie, Class D amplifiers. THD & IMD figures at normal output levels for many of them fall far below anything that's going to be audible to anyone anyway, so they're not necessarily any more of a weak link than an entirely analogue design.

[Slipster]

@Slipster, I'll have to take issue with you re:ABX vs. 16/44. Ear/Brain doesn't just work in Amplitude+Frequency domains but also in Time domain. In this respect, 24/96 or 24/192 IS clearly & noticeably better, and I can tell the difference with phase information (placement) in Stereo/Surround programs and in the reverb tails. Once you're trained to work with those particular variables, the difference isn't minute. Otherwise, what you say is quite valid.

Agreed. Training can lead to an increased probability of spotting anomalies that the untrained listener won't detect, but I still take issue with comments along the "clearly & noticeably better" line as this only applies to those who've trained themselves meaning that even the tiniest of discrepancies can seem to jump out and bite you on the nose.

What percentage of listeners do you think train themselves specifically to notice artifacts? Vastly below 0.1% would be a safe guess, IMO. Statistically, Mr Average often has a tough time discriminating a LAME MP3 encoding at -V6 (~115kbps) from 44.1/16 lossless in my experience, and the effects of that are several orders of magnitude more destructive than going from 192/24 to 44.1/16.

On a side note, it's very easy to make a total pig's ear of converting 192/24 or 96/24 down to 44.1/16, and it's the conversion process that nearly always leads to ABX tests of such material being shot down in flames as further analysis clearly shows in nearly every case that the 44.1/16 output is nowhere near as close to the higher bitrate/bit-depth reference as it could have been if the conversion had been carried out by a competent technician using decent software. Such tests then become less of a case of comparing different varieties of apple and more a case of comparing an apple to a lemon.

Me, I'd still use 24/96 or 24/192 to master with - that way degradation can occur but never build up to the point of messing up transparency. For most end uses, 16/44 or 16/48, or maybe 20/48 (with the correct high quality chain) are plenty good.

I don't see any logical reason to not use 32-bit float at every stage pre-final master (assuming that all of the effects plug-ins support it), but I've yet to see results for an ABX listening test with a sufficiently high confidence level to suggest that properly encoded 44.1/16 for stereo material isn't adequate for everyone at the delivery end.

[Cornucopia]

Ok, I see where you're coming from.

Edit: though I would give the average Joe a little more credit, re: -v6mp3. At least as long as they were listening on something better than their ipods through earbuds on a busy street.

Scott

[pandy]

I don't see any logical reason to not use 32-bit float at every stage pre-final master

32 bit float is not enough - in many cases 48 - 56 bit INT is sufficient to provide accurate calculation in audio world.
You probably disagree with this and ask why?
Check IIR filters requirements = 24 bit resolution for particularly high order IIR filters is not enough - 64 bit float seems to be to expensive (power and silicon size) and this is why companies like Analog Devices, Texas Instruments use in their Audio DSP CPU's combination like INT 28 - 80 bit data path.

[Slipster]

32 bit float is not enough - in many cases 48 - 56 bit INT is sufficient to provide accurate calculation in audio world.

Wow! I was aware that some IIR filters were using 64-bit float internally to improve the precision of calculations, but are they using 64-bit float externally too?

[pandy]

32 bit float is not enough - in many cases 48 - 56 bit INT is sufficient to provide accurate calculation in audio world.

Wow! I was aware that some IIR filters were using 64-bit float internally to improve the precision of calculations, but are they using 64-bit float externally too?

My point was that 32 float not provide to many advantages over 24 bit INT and it is worse that 32 INT when algorithm is designed to work in INT world.
32 float is easier for developers - they can be more PC programmers than DSP coders so at some point float promote careless way of writing code than INT.

[hello_hello]

Tests still occur as one-offs on an almost weekly basis on another forum I visit occasionally when people turn up saying "MP3 is crap!" only to be asked to take an ABX test and, upon taking a test, fail it quite convincingly. The sad fact is that so many turn up with the "MP3 is crap!" attitude and refuse to take a test, thus carrying on to spread the message with no practical evidence on their part to support their claim.

It never ceases to amaze me.... when some people have made up their minds, they've made up their minds, and no evidence to the contrary will ever change that.
I recall getting into quite a lengthy MP3 debate in another forum a few years ago. I went to the trouble of starting a new thread and uploading some test samples.... and the "MP3 is crap" brigade also turned that thread into a mess quite quickly, without a single one of them being willing to download the samples and take the test.

By the way.... having uploaded A/B files myself, I worked out how to cheat pretty quickly. I can pick which is the original and which has been converted from MP3 without needing to bother with all that listening to them fuss. Should I tell??

[Slipster]

By the way.... having uploaded A/B files myself, I worked out how to cheat pretty quickly. I can pick which is the original and which has been converted from MP3 without needing to bother with all that listening to them fuss. Should I tell??

Shhhhh! Don't give them any more ammo for gawd's sake!

PS An irrational debate has broken out in the "CD to MP3" thread now where claims are being made that a certain well known lossless format isn't lossless, and that it's also impossible to make perceptually transparent MP3 encodings. I'm done there, but you might want to take a look.

[FulciLives]

This thread is a disaster.

[turk690]

This thread is a disaster.

lol