Recorded CD sound quality

[Rob Babcock]

One of my favorite CDs is a classical recording that was done in digital in the mid 80's. Supposedly this should sound gruesome, but if it does I've never noticed it. Could just be that it's one of my favorite works.

 

[WmAx]

Doesn teh ABX software allow for performing the comparison on my computer or would it mean I'd have to connect my PC to my main rig? My PC sound rig is pretty modest: Griffin Powerwave Tripath amp/soundcard driving a pair of JBL HLS-610s. Fine for cranking tunes while I surf or monitor compilations, but hardly high end.

I have no knowledg of the speakers you refer, nor do know what set up/positiong you are using. If your PC soundcard is low noise, your computer is not noisy(fans, etc.) and you don't end up with any ground loops causing noise when you connect to the main rig, it will be satisfactory. Just need to get teh monitor and keyboard in a position where you can comfortably sit in the sweet spot for the comparision. A high quality reference headphone connected to the cmoputer would be more practical and more sensative to artifacts.

-Chris

 

[WmAx]

That would be interesting. I'll have to take some time, though- it's been a while since I've used MP3. I'll have to see what I can find. If memory serves, between the S&V article and my own admittedly subjective recollections, 1) massed strings, 2) cello, 3) a large brass section, 4) piano & 5) large orchestra were the least satisfying things I've encoded. Again, I may not have used the best encoder, either. Lastly, some of my playback gear doesn't like MP3: my older Pioneer won't always play them & sometimes has static, and my Denon doesn't always recognize any track #'d above 20. I have no idea why.

PM me your addy and I'll try to come up with a CD-R of wav files. Be advised that it may take me a while to do so, though.

INdeed, you seem to have not much use for compressed audio format. But as I have stated in other threads, I use a portable player often and I need to use a compressed format. I also listen to different music on a portable as opposed to my home system. Portably, I tend to listen to a lot of pop type music as opposed to my home system where I prefer to listen to acoustic, jazz and classical recordings.

This little 'fun' experiment will be for just that -- fun. I have sent my address to you via PM as you requested.

-Chris

 

[av_phile]

The 'poor' quality of lossy compression is essentially a myth. Sure, with poor encoders and/or insufficient bitrates for a given codec, yes, their willl be problems. But with hi quality encoder(s) and the proper bitrates, even picking out the lossy compressions on a very high quality reference headphone(more sensitive then any speakers due to no room effects/masking) is exceedingly difficult, and limited to particular 'problem' sound samples that are not represenative of the overwhelming majority of the music program. But if you are listening and cmoparing in NON-CONTROLLED circumstances, then it may seem 'obviously' inferior. This evidence suggest that this is a subconscience bias, not a reality.

-Chris

Well I did say, rightly or wrongly, you can have near CD quality on MP3 for SOME materials at high bitrates. They are mostly pop music with very compressed originals to begin with. For the most part, you can't distinguish unless you listen hard enough.

When I manipulated some cd tracks on my PC, converting them to MP3s at 192kbps, I did notice a visual confirmation that the waveforms as shown on the NERO software between the wav and mpr3 are DIFFERENT in marginal but distinct ways. Playing these two materials sequentially did not exhibit any marked difference for me for that particular track. But I wasn't interested in hearing any difference as I had some urgent reasons to convert wav files to mp3s.

However, when I experimented later and converted the MP3 back to wav, then back to MP3, then again to WAV, it appeared the codecs did change the character of the file at each conversion to MP3 so that the end WAV file was a lot more different than any of the previous files, especially the first wav file. Transcoding to MP3 did materially alter the file so that subsequent transcoding made the final product look different, like multi-generation tape copies, though not as severe. I did not experience the same thing when copying wav files as another wav file, no matter how many times I did (though some other purists will argue that as well) That's enough to convince me what other audiophiles have long asserted: MP3s are not for serious listening.

I don't see anything mythical about the technical aspects of Lossy compression that throws away pressumably unnecessary signals in their codecs to conform with psychoaccoustical theories. As the type of compression suggests, it losses some signals. That's a technical reality. The fact that, as you yourself said, there can be a "problem" with some rare sound materials and that discerning on excellent headphones can be "difficult" but not impossible doesn't make the PERCEPTION a myth. The fact that the original signal has been MANIPULATED where some signals are irretrievably lost buttresses the perception against MP3 files from a pure high-fidelity standpoint.

 

[WmAx]

Well I did say, rightly or wrongly, you can have near CD quality on MP3 for SOME materials at high bitrates. They are mostly pop music with very compressed originals to begin with. For the most part, you can't distinguish unless you listen hard enough.

Don't confuse dynamic compression with data compression. I'm sure you do not mean to make such an erroneoius correlation.

When I manipulated some cd tracks on my PC, converting them to MP3s at 192kbps, I did notice a visual confirmation that the waveforms as shown on the NERO software between the wav and mpr3 are DIFFERENT in marginal but distinct ways.

Well, it is lossy compression. I must have some differences - but the question is of audibility, not of visual/meaurable difference alone.

However, when I experimented later and converted the MP3 back to wav, then back to MP3, then again to WAV, it appeared the codecs did change the character of the file at each conversion to MP3 so that the end WAV file was a lot more different than any of the previous files, especially the first wav file.

Of course this caused substantial problems. This is not suprising in the least.

Transcoding to MP3 did materially alter the file so that subsequent transcoding made the final product look different,

Thus, why transcoding is never recommended for lossy compression formats.

That's enough to convince me what other audiophiles have long asserted: MP3s are not for serious listening.

How is this logical? If the compressed audio is not discernable from the original in a DBT -- how is it not for serious listening? If it's a problem sample/example that can not be encoded properly(not common), then of course the codec would not be useful for serious listening for that example. Perhaps you are referring to an issue of reliability -- please see next paragraph...

That's a technical reality. The fact that, as you yourself said, there can be a "problem" with some rare sound materials and that discerning on excellent headphones can be "difficult" but not impossible doesn't make the PERCEPTION a myth. The fact that the original signal has been MANIPULATED where some signals are irretrievably lost buttresses the perception against MP3 files from a pure high-fidelity standpoint.

I guess this depends on one's purpose /goal. If by 'pure hi-fidelity' you mean that all samples are perfectly reproduced transparently 100 percent of the time, then you are surely correct. But no one is suggesting(from what i can tell) that mp3, for example, is 'pure hi-fidelity'. I personally use the format for portable use only. The conversation was directed at teh implication that mp3 was only suitable for pop music as you originally claimed. Regardless of a portable or 'hi end' speaker system, mp3s made with a sufficient encoder and at a sufficient bitrate should remain transparent in most circumstances.

-Chris

 

[av_phile]

Don't confuse dynamic compression with data compression. I'm sure you do not mean to make such an erroneoius correlation.

Nope I didn't mean that. Pop music employ a lot of dynamic compression, more than jazz or classical music. But it appears to me that dynamically compressed materials lend itself more faithfully to data compression. It seems that the "masking" effect in psychoacoustics applies more generously in such materials so these masked signals can be dispensed with and not be entirely audible in data compression. But feel free to correct me on this.

Well, it is lossy compression. I must have some differences - but the question is of audibility, not of visual/meaurable difference alone.

It didn't strike me as audible when I first dabbled in MP3. Though the visuals on the NERO did make me uncomfortably suspicious.

How is this logical? If the compressed audio is not discernable from the original in a DBT -- how is it not for serious listening? If it's a problem sample/example that can not be encoded properly(not common), then of course the codec would not be useful for serious listening for that example. Perhaps you are referring to an issue of reliability -- please see next paragraph...

You may not have been long enough in this hobby to notice that most of the things audiophiles in this hobby do and don't can't ever be called logical.

I guess this depends on one's purpose /goal. If by 'pure hi-fidelity' you mean that all samples are perfectly reproduced transparently 100 percent of the time, then you are surely correct. But no one is suggesting(from what i can tell) that mp3, for example, is 'pure hi-fidelity'. I personally use the format for portable use only. The conversation was directed at teh implication that mp3 was only suitable for pop music as you originally claimed. Regardless of a portable or 'hi end' speaker system, mp3s made with a sufficient encoder and at a sufficient bitrate should remain transparent in most circumstances.

-Chris

No argument there. My goal has always been high fidelity. For portability, MP3s are best. Though there are newer formats coming out promising better "fidelity" like MP4 or WMA. I would question the need for such, as many of the portable playback gears are themselves often wanting in the Hi-fidelity department.

And I don't claim mp3 to be suitable to pop "only." See the earlier part of this post.

 

[realistic]

Actually its the opposite. Highly compressed material (dynamic compression) is a poor source for the encoder. Take a track from a cd that is very compressed, but still not clipping, and encode it into mp3. The result will almost invariably clip on playback.

The definition of digital clipping varies, but most conservative software uses 4 consecutive max value samples to be clipping. That is usually not audible (4/44,100 secs) but highly compressed sources result in many more than 4 consecutive max value samples and can become audible.

 

[mtrycrafts]

and I must say that with the few titles I have, I can't distinguish the SACD stereo tracks reworked in DSD from the early CDs.

.

Clear indication that it is not the CD medium that is inferior to SACD or DVD-A but the recording engineering of todays CDs.

 

[Rob Babcock]

I don't own a pair of headphones- I really should get one, as the Griffin Pwave is supposedly a very good headphone amp. My JBL speakers are probably 4-5 years old, and pretty darned decent for mass market cheapies (Corey Greenberg raved them in one of the last issues of Audio magazine- I really miss them).

Should be interesting. I can definately see the appeal of MP3 for portable use, and perhaps one day I'll break down & finally buy Zen Nomad; I've kinda had my eye on the 40 GB version (it works as a portable hard drive for moving data, and that would be extremely handy for me).

I can make no great claims to possessing a Golden Ear. It should be interesting to see how high the bitrate has to be before it's transparent. My 128 kbps attempts (ostensibly CD quality) have sounded fairly grim, but again, maybe I did something wrong. My 'speriments, using friends and family as guinea pigs, revealed that my MP3s had to be about 256 kbps before they couldn't reliably tell them from the CD with rock.

I would be interested in eventually hooking a PC up to my main rig, but I don't have a spare PC right now. That would be great for ABX tests.

 

[Rob Babcock]

Clear indication that it is not the CD medium that is inferior to SACD or DVD-A but the recording engineering of todays CDs.

I think CDs greatest failing is that it's 2 CH only. While I'm not yet completely convinced there's no benefit to going with higher resolution, I've long felt that the major advantage the newer format has is MC sound.

 

[WmAx]

If your interested, some headphones that are popular and considered of high quality by so-called audiophiles:

Sennheiser HD-580($150)
Sennheiser HD-600(essentially the same as HD-580 but with slight different cosmetics and higher price)($200?)
Sennheiser HD-650($350)
Grado RS-1($600?)
Grado RS-2($500?)
Sony MDR-CD3000($400)
Sony MDR-CD900ST($180)
Audio Technical ATH-900($190)

These are just a few recommendations, and certainly not complete list. Prices are ballpark average online prices that I remember -- not going to look up each one for this! Prices do not nessacarilly represent a correlation in sound quality.

Concerning the bitrate for transparency -- as I've said before; it depends on the encoder, the setting used in that encoder and the source material. It's certainly no suprise that 128kbps was not transparent! I don't think any format is transaprent at that rate -- that I am immediately aware.

BTW, check out www.allofmp3.com . I just noticed this site a little while ago, and they offer different encoder rates for the music as well as lossless formats for some newer releases.

-Chris

I don't own a pair of headphones- I really should get one, as the Griffin Pwave is supposedly a very good headphone amp. My JBL speakers are probably 4-5 years old, and pretty darned decent for mass market cheapies (Corey Greenberg raved them in one of the last issues of Audio magazine- I really miss them).

Should be interesting. I can definately see the appeal of MP3 for portable use, and perhaps one day I'll break down & finally buy Zen Nomad; I've kinda had my eye on the 40 GB version (it works as a portable hard drive for moving data, and that would be extremely handy for me).

I can make no great claims to possessing a Golden Ear. It should be interesting to see how high the bitrate has to be before it's transparent. My 128 kbps attempts (ostensibly CD quality) have sounded fairly grim, but again, maybe I did something wrong. My 'speriments, using friends and family as guinea pigs, revealed that my MP3s had to be about 256 kbps before they couldn't reliably tell them from the CD with rock.

I would be interested in eventually hooking a PC up to my main rig, but I don't have a spare PC right now. That would be great for ABX tests.

 

[WmAx]

I think CDs greatest failing is that it's 2 CH only. While I'm not yet completely convinced there's no benefit to going with higher resolution, I've long felt that the major advantage the newer format has is MC sound.

Again, how do you approach this from the 'i'm not convinced there's no benefit to going to highe resolution' perspective?

When was it proven to be beneifical in the 1st place?

-Chris

 

[WmAx]

Nope I didn't mean that. Pop music employ a lot of dynamic compression, more than jazz or classical music. But it appears to me that dynamically compressed materials lend itself more faithfully to data compression. It seems that the "masking" effect in psychoacoustics applies more generously in such materials so these masked signals can be dispensed with and not be entirely audible in data compression. But feel free to correct me on this.

I'm not aware of any such condition (dynamic vs. less dynamic) having an effect of significance on the encoding effectiveness.

You may not have been long enough in this hobby to notice that most of the things audiophiles in this hobby do and don't can't ever be called logical.

I've been in the hobby long enough. As odd as it may sound; I used to be on the subjectivist 'side'. I believe I understand 'why' that side exists.

No argument there. My goal has always been high fidelity. For portability, MP3s are best. Though there are newer formats coming out promising better "fidelity" like MP4 or WMA. I would question the need for such, as many of the portable playback gears are themselves often wanting in the Hi-fidelity department.

I am not aware of WMA(at least not the normal variant that is widespread) demonstrating better fidelity at a given bitrate then a good mp3 encoder. As for mp4, that is supposedly slightly better at a given bitrate. However, with 20 Gig and higher being the norm on those HD based mp3 players, mp3 is just fine IMO at 256kbps or higher with a high quality encoder. Lower bitrates will usually be transparent on much material(192kbps) but I prefer to leave a bit of headroom so that fidelity is marginally more reliable.

And I don't claim mp3 to be suitable to pop "only." See the earlier part of this post.

Perhaps. But I drew this conclusion based on this statement from you: "I've heard them and they do offer respectable quality with near CD quality at high bitrates for some compressed pop materials. "

-Chris

 

[Rob Babcock]

We've discussed that topic ad nauseum elsewhere, Chris, and no offense, but I don't intend to revisit it here. You cited some evicence, most of which was very old or completely inaccessible to me (ie no link or way to read it online). I'm not implying you made it up or anything, but I'm a skeptical person, and the mountain of evidence isn't there. There may in fact be no benefit to higher rates, but I'm not convinced yet either way.

You may in fact be correct, but there's no reason not to go with higher bit rates nowadays- 24 bit gear is common and storage is exceedingly cheap. IMOHO, having higher resolution will leave more margin for error- I have no evidence of this, but I speculate that if there's a difference in the sound of DACs it would be how accurately they reproduce the LSB. It would seem to me (again, based on reasoning, not evidence) that a cheap 24 bit DAC should offer better performance than a cheap 16 bit DAC (assuming 24 bit & 16 bit recordings, respectively). Therefore, I'd expect a cheap (in the engineering sense, not price) DVD-A player to sound better than a comparable Redbook CD player.

I'm not an engineer, and my conjectures therefore might be quite dubious- hell, they might be laughable. I'm sure you'll no doubt inform me if they are. But on paper, more bits & a higher sampling rate will capture more resolution, and until I'm convinced beyond a shadow of a doubt that no one is capable of hearing that information, I'd prefer we keep it. That's not reasonable, perhaps, but I think it's the most conservative approach.

 

[WmAx]

If you wish to re-visit those threads, I offered to email you any paper you wished to view. It's irrelevant how old the papers are, that simply is not a factor unless someone shows a fatal error in those tests that is relevant. The papers are generally not avaialbe freely online. I think only one was avaiable for free. JAES, for example, typically charges a fee for each downloaded paper except for the ones in their spec section.

-Chris

We've discussed that topic ad nauseum elsewhere, Chris, and no offense, but I don't intend to revisit it here. You cited some evicence, most of which was very old or completely inaccessible to me (ie no link or way to read it online). I'm not implying you made it up or anything, but I'm a skeptical person, and the mountain of evidence isn't there. There may in fact be no benefit to higher rates, but I'm not convinced yet either way.

You may in fact be correct, but there's no reason not to go with higher bit rates nowadays- 24 bit gear is common and storage is exceedingly cheap. IMOHO, having higher resolution will leave more margin for error- I have no evidence of this, but I speculate that if there's a difference in the sound of DACs it would be how accurately they reproduce the LSB. It would seem to me (again, based on reasoning, not evidence) that a cheap 24 bit DAC should offer better performance than a cheap 16 bit DAC (assuming 24 bit & 16 bit recordings, respectively). Therefore, I'd expect a cheap (in the engineering sense, not price) DVD-A player to sound better than a comparable Redbook CD player.

I'm not an engineer, and my conjectures therefore might be quite dubious- hell, they might be laughable. I'm sure you'll no doubt inform me if they are. But on paper, more bits & a higher sampling rate will capture more resolution, and until I'm convinced beyond a shadow of a doubt that no one is capable of hearing that information, I'd prefer we keep it. That's not reasonable, perhaps, but I think it's the most conservative approach.

 

[WmAx]

but there's no reason not to go with higher bit rates nowadays- 24 bit gear is common and storage is exceedingly cheap.

It's true that 24 bit systems can be dirt cheap today. But primarily I like to discuss the possibilities/limitations, not the cost. It's just more fun.

-Chris

 

[Rob Babcock]

I'm not doubting you, and certainly age alone wouldn't invalidate the results, especially if you mean analog tape vs 16 bit PCM.

I have a logical bent, but I just feel a bit more comfortable with the extra resolution of 24 bit. I'm agnostic about DSD- I haven't seen much research comparing it to PCM aside from Sony adspeak. My own ears tell me DSD offers no improvement over PCM, but my experience also is just anecdotal. The best sounding recordings I've ever heard personally were PCM, but obviously it's early in the "game" to really say much.

There's an obvious difference between the "16 bit vs 24 bit" issue and, say, "$5 cable vs $1500 cable". In the case of the former there's a measureable performance difference- the issue is whether that difference is audible to a human being. With the latter, the issue is whether there's any genuine difference there at all. That's an important distinction to me, and that's why I also find discussing digital more fun.

 

[av_phile]

Actually its the opposite. Highly compressed material (dynamic compression) is a poor source for the encoder. Take a track from a cd that is very compressed, but still not clipping, and encode it into mp3. The result will almost invariably clip on playback.

The definition of digital clipping varies, but most conservative software uses 4 consecutive max value samples to be clipping. That is usually not audible (4/44,100 secs) but highly compressed sources result in many more than 4 consecutive max value samples and can become audible.

You may be right, I'm no digital expert to either agree or not. My thoughts were purely a product of a mental exercise that goes this way:

If there are 5 instruments playing with varying degress of loudness as is often the case in pop, there will be moments when some or most of the instruments can MASK the rest, so those MASKED signals can be safely discarded at that time by a lossy compression recording, since, upon playback, the human ear would not have to hear them anyway. And because they're dynamically compressed. meaning you only have on level of loudness and weakness all throughout a track, then the compression engine would have a much easier time.

However, in a classical orchestral track where there is a solo flute or bassoon playing a pianissmo passage, there's really no other instrument to mask (except the silence) and hence the lossy compression may not apply its algorythm at that passage and let the solo passage go through untouched, expect the silent moments which are so easy to compress. And because a classical piece would have varying moments of loudness and weakness, the compression engine would be making more mathematical processes that just might result in errors.

But that's just a layman's thinking tyring to undersand this compression business. So if what you say is right, then I stand corrected in my assumptions.

 

[jjwinc]

Hi,

I had a question. I'm not an expert, but over the last couple of weeks I was auditioning new speakers, and when listening to the higher mid-range brands (mostly Paradigms) I noticed a distinct difference in sound quality with older CDs versus newer ones. I was told that there are several factors that will contribute to this, but the bottom line is the better quality going into the recording process, the better sound. That's why bands like Steely Dan and new more advanced recordings sound so good, they take the time to do it right.

So, my question is, when people flock to music downloading sites, how do these burned CDs sound on a high quality system with good speakers? Do people who download music simply not care about sound quality? I know a lot of people who listen to music solely on their computers.

Either way, just wondering. Thanks and have a nice day.

phred

Most of the earlier recorded cd,s sound terrible on high end systems.I own the paradigm studio 100v2. some of my older cd,s sound like crap played on these speakers especially the ones recorded from analog to digital in the 80,s.Guess this shows how far digital recording has come in the last 15 yrs.