A discussion of ripping and encoding/decoding software

[bob p]

When you say "digital audio" I believe that you are referring to files that have been molested by lossy compression (namely MP3 files). I personally find unmolested digital audio files to have much more potential than my scratched up vinyl albums. While cds are 16 bit/44.1mHz, you could make digital audio files that are 24 bit and 96mHz for much higher resolution and bandwidth.

I've converted a lot of my scratched up albums into CDs, and the results have been really good- once I find and correct the flaws that can be corrected, I never have to worry about it getting more scratches (as I would with a vinyl album).

But there is a downside to the AD conversion process used in making CD's- as Speed Racer used to say, it can make the drummer's cymbals sound like a salt shaker...

I have to say that the more recent CD's usually have decent cymbal sounds- perhaps they sample the tracks at 24 or 32 bit for the initial conversion. Or maybe they just try to avoid the annoying frequencies when recording and mixing it down.

I think of MP3's as having the same fidelity as radio broadcasts and audio cassette recordings when I was growning up- certainly nothing near as good as a long play record. I don't understand the point of downloading an album in MP3 format, unless you just want to audition it in lo-fi to help you decide if you should get a *decent* copy of it later. LOL

I have to admit, I'm still not enamored with any of the digitized recordings, for reasons like those that Speedy mentioned -- proper quantization of some signals just isn't as easy as most people think it is. The sounds of cymbals, glockenspiels, triangles, etc. can be particularly revealing tests for critical listening. As it turns out, when I had written earlier about the A/B tests between the $700 Denon and the $1000 Nakamichi CD players, I was listening to classical recordings that had triangles and glocks in them. In some respects, having a good ear can be a curse. HiFi audio is alot like wine. Once you can taste the difference between a $130 and a $150 bottle of wine, you're cursed for life! Especially when you have a beer budget!

I guess I would agree that the MP3 is the new "demo" format that many people will use to decide if they want to own the track on "hifi" media or not. The recording industry continues to see sales growing year over year, so I guess digital piracy isn't really hurting them. I guess Lars whats-his-name from Metallica successfully intimidated a lot of customers with those Napter lawsuits against the 12-year old kids.

Regarding radio, analog broadcast radio can be very HiFi, even though alot of the time it isn't -- the problem is that most stations just rolled off everything above 15K for whatever reason. As an aside, I've acquired an 1964 Scott 350-D Stereo Tuner (last of the all valve tuners from Scott), which was regarded as the premier tuner of the tube era. It needs tubes, but believe it or not, got it in a dumpster dive and saved it from going into the landfill. Now if I could find a free McIntosh or Dynaco power amp on its way to the landfill...

[David Schwab]

I have to admit, I'm still not enamored with any of the digitized recordings, for reasons like those that Speedy mentioned -- proper quantization of some signals just isn't as easy as most people think it is. The sounds of cymbals, glockenspiels, triangles, etc. can be particularly revealing tests for critical listening.

I think another issue is the brickwall filter used to prevent aliasing. There is information above our hearing that we use, such as phase relationships and stuff. We lose most of that with standard digital formats. The upper harmonics aren't accurate. The resolution is too low.

I think this DVD audio format should improve the situation greatly, since you will be able to use a higher sample rate and bit depth.

Still doesn't help with lossy compressed audio though.

[Steve Conner]

I don't get it. Surely you can either hear a frequency or not, and if you can't hear it, it doesn't matter if it's filtered out.

Of course, it is possible for filters to cause weird phase shifts and amplitude ripples in their passbands, so a filter that cuts at 20kHz could be messing with phase all through the teens of kHz. It's very difficult to design brickwall filters that let the whole audio band through, stop all aliasing from a 44.1kHz sample rate, and don't ripple either the amplitude or the phase of signals near the top of the passband: a tradeoff has to be made between all of these quantities, and the "sound" of the DAC depends on the particular balance the designer chose.

I think that is the sole reason for audible differences between DACs, and the presence of ultrasonic information is a red herring. It must be, or else everyone who ever measured the range of human hearing got it wrong.

Personally, I think I prefer the sound of DACs that use a less vicious filter to save mangling the HF phase and transient information, at the cost of either rolling off a little of the extreme high end, or allowing a bit of aliasing to bleed back into the passband. Some DIY DACs take this to an extreme and use no anti-alias filter whatsoever, either digital or analog.

[David Schwab]

I think the whole idea is we do not perceive frequencies above a certain point as "sound" but there is still information there that we process. Same can be true of very low frequencies.

Our brains process a whole lot of stuff that we are consciously unaware of.

Also a very severe brick wall filter has to be doing something to the phase alignment in the upper harmonics, and that can't be a good thing.

I'd imagine at some point that storage will be so cheap, and internet connections so fast, that we will be able to use uncompressed audio in our portable devises. Or at least non lossy compressed files.

[Steve A.]

I have to admit, I'm still not enamored with any of the digitized recordings, for reasons like those that Speedy mentioned -- proper quantization of some signals just isn't as easy as most people think it is. The sounds of cymbals, glockenspiels, triangles, etc. can be particularly revealing tests for critical listening...

Bob:

I guess you are looking at things as an audiophile while I look at them as a someone who has been making serious home recordings for 30 years (figure 40 years for some not-so-serious stuff). I would work hard mastering 2 or 3 tracks to a 7 1/2 ips reel-to-reel tape and then try to dub decent copies to audio cassette, hoping that they would play right on everybody's tape deck. 22 years I got my first 4 track home studio, which upped the ante a little bit, but was still stuck with using audio cassettes for distribution. So being able to edit digital files and burn them to a CD, or distribute them as FLAC files on the internet- that is so much better than things used to be!

I don't think I hear much over 10khz anyway, and if I get cymbals sounding like a salt shaker, I either take them out of the mix, or just EQ so that they sound like something altogether different... LOL

As for dubbing LP's to CD, it sure beats dubbing them to audio cassettes! I just did an album that I had picked up 30 years ago as a cut-out and it had hardly been played at all (Mel Brown's The Wizard on Impulse!) and it sounds great to me. I polished up the digital files and it sounds better than it ever did on a turntable (since I removed the few pops and clicks). As a matter of fact I just uploaded an MP3 of it as a sound sample:

Mel Brown - Ode To Billie Joe [Impulse! A-9169 - 1968]

http://www.megaupload.com/?d=0C71TNC4

Steve Ahola

[Steve A.]

I had a big discussion with someone about this a couple of years ago. All audio CD's have error correction built in (Cross-Interleaved Reed-Solomon Coding), because during the manufacturing process, there are going to be physical defects in the aluminum that holds the pits. You end up with a combination of random and burst errors.

Any time you play a CD, there is going to be some bits rounded off and so on, and you really can't hear the difference. A CD player uses error correction to play back the CD. A program like EAC claims to do an exact copy, but all it's doing it making an exact copy of a disk that already has errors! Only it's using the error correction algorithm that is built into every CD player. This is a tad redundant, since playing that exact copy of the "flawed" CD will be corrected at the CD player.

So first the CD player uses an interleaving scheme, and if that doesn't work it uses an interpolation scheme where it tries to "guess" the content of the sample by looking at the nearest neighbors. it wont "fix" the error, but it will make it inaudible by gracefully degrading the signal to mask the clicks and pops.

iTunes has an error correction setting, that uses the same principals as EAC to copy an audio CD.

So in my opinion, any quality CD duplicating program will make an exact copy of an audio CD... which is not to say the CD is not flawed in the first place. Things like EAC might help with damaged CD's, but I've never heard an improvement in sound quality...

One of the reasons why EAC is the recommended or required ripper at many sites is that is does generate rather detailed reports, noting your settings along with any errors in the process. EAC also supports AccurateRip, which compares the results of your rip with a rather large database, giving you a confidence rating (AccurateRip will also do an initial test of your drive to help you set the read offset value properly).

Error correction or not, sector alignment errors (aka "sector boundary errors") can produce noticeable pops between tracks which were not on the original CD. (This is only noticeable on albums in which the tracks run together, such as live albums or a very obscure album evidently for astonomy buffs called "Dark Side Of The Moon".)

You bring up a good point- it would be interesting to rip the same CD several times in EAC to see if the checksums are the same. And then try other programs to see how the checksums compare.

As you suggest, *EXACT* Audio Copy might be a bit of a misnomer, but it does seem to generate more consistent results than most of the other programs out there. (dbPowerAMP is supposed to be coming out with a new version with more bells and whistles than EAC.)

The other programs seem to work okay if you intention is to create a library of MP3 files, but if you will be burning them to CD I think that you will get better, more consistent results with EAC. Other than clicks from sector boundary errors, I'm not sure if any of us can really tell the difference in listening to a single audio file ripped from a CD using the different programa (I think that the read offset has to do with when the track starts and ends, and doesn't affect the actual audio data in the middle).

Some ripping programs automatically strip out the silence at the beginning and end of each track, which can affect the sequencing if you burn them to disc. In a well-produced album, the gaps between the songs are like white spaces in a painting. If you want to have the same experience as the original CD, the gaps should be the same, as well as the relative volume between the tracks (like if someone decides to "normalize" all of the tracks individually... a definite "no no"... LOL)

Steve Ahola