Columns Retired Columns & Blogs |
Loudspeakers Amplification | Digital Sources Analog Sources Featured | Accessories Music |
Columns Retired Columns & Blogs |
Loudspeakers Amplification Digital Sources | Analog Sources Accessories Featured | Music Columns Retired Columns | Show Reports | Features Latest News Community | Resources Subscriptions |
Are you saying you managed to produce mp3s that sound better than cd ?
Better-than-CD? Yes, as audacious as that sounds. But don't take my word for it. Please visit the links in my orginal post above, give a listen, and judge for yourself.
Having spent over 30 years in live performances, I've always been disturbed by the many qualities lacking in CD playback. It's not a matter of how much you spend on your stereo; a great deal of the information required for authentic playback is simply missing. Rather, it's actually resident within the recordings, but indecipherable and unavailable for playback.
I've found it absurd that in codec development everyone is striving for 'lossless' or 'CD-quality' as a standard. Frankly, CD is a terrible disappointment to anyone accustomed to the dynamics of live performances. My standard is the actual live performance itself rather than CD, and restoring that magic in compressed formats has been my goal.
So please give a listen, and decide for yourself. Your feedback is most welcome here!
Regards to all Stereophiles,
Tuvia Wertheim, Compaxx Producer & Stationmaster
email: stationmaster@compaxx.org
Thank you for making comparisons available. The Compaxx versions were made from the original CD, correct? Not from the MP3 versions that you label as "CD"
I do have a question about your EULA on the Compaxx comparisons page. You write "Modification of the materials or use of them for any other purposes other for personal evaluation by the original recipient of said materials is a violation of the copyrights and other proprietary rights."
What exactly do you mean by this? If I convert the Compaxx files to WAV format in order to do some data analysis, does this infringe the EULA? If I publish my findings in Stereophile magazine or on this website, does that still count as "personal evaluation"? I just want to make sure I am not going to inadvertently infringe your conditions.
John Atkinson
Editor, Stereophile
Thank you for your listen, John. Believe it or not, the Compaxx-remastered files were actually made from the mp3 files labelled as 'CD' that precede them. The forensic data-mining technique used in Compaxx is capable of restoring even the 'lost' data from mp3. The difference in terms of results between this method as opposed to applying Compaxx to the original PCM-wav would be extremely slight at the bitrate used (192k, stereo). For the sake of a fair comparison I've presented in this fashion as stated on the webpage: "ALL COMPAXX FILES HAVE BEEN MADE DIRECTLY FROM THE ACCOMPANYING CD FILES THAT PRECEDE THEM." (In my production work, I prefer to work wherever possible directly from CDs for optimal restoration.)
> Regarding the Compaxx EULA: You may freely torture-test the downloadable samples for evaluation purposes. It is forbidden however, to retag or redistribute these files without permission (such as web samples for promoting the sale of CDs). I'd be delighted if you were to review these if you wish, and you may certainly reference them freely both here as well as in print.
I appreciate your asking about this, John. We're all cautious these days about legal issues. I do the best I can to always remain DMCA and RIAA compliant in my work, and respect how trying this must be on an audio/stereo website such as "Stereophile".
> Decoding to WAV for analysis: You're welcome to try this, but bear in mind that a significant amount of the embedded subcode risks damage in the process (in particular the spatial cuing and subwoofer subtracks). For best results I'd recommend analog-playback analysis, or digital i/o reads taken directly off the decoder chipset.
I also have these samples on both CDROM, CD, and CD-Extra for laboratory testing and will forward to Stereophile if you wish. An 8 1/2 hour Feature-length Compaxx Masterworks Classical mp3 disc is also available upon request.
If you have any further questions or comments, please feel free to post them here or contact me directly. Your feedback is most welcome.
Tuvia Wertheim, Compaxx Producer & Stationmaster
Okay. Thanks for getting back to me.
Okay. I downloaded 3 sets of files to compare, the Jennifer Warnes & Joe Cocker track, Joni Mitchell's "Turn Me on, I'm a Radio," and the Four Seasons excerpt. In each case, the Compaxx version sounded more vivid and more detailed. However, it also sounded both brighter and significantly louder.
I therefore looked at the recorded spectrum file statistics using Adobe Audition. Using the Four Seasons as an example, both the peak level of the file and the average level had been increased by the Compaxx process. The peak levels of the original MP3 were -0.26dB (L) and -5.13dB (R) compared with +0.1dB (L) and -0.21dB (R) for the Compaxx version, which also had 4 instances of clipping. The difference in the average RMS levels was more extreme, at -22.77dB (L) -24.92dB (R) for the original, -10.73dB (L) and -12.98dB (R).
I don't know what Compaxx is doing that is proprietary, but this analysis suggests that the effect of the low-level information reconstruction algorithm, if it exists, is overshadowed by simultaneous use of a conventional multiband compressor of some kind.
I did find the Four Seasons to sound more reverberant, but this may be an side effect of the compression used rather than the use of an artificial reverb processor of some kind.
Summing up, while the Compaxx processing may be beneficial for those listening to a low bit-rate stream on typical desktop computer speakers, I don't believe, at least on the evidence of these three music examples, that it represents a major breakthrough in signal processing or information retrieval. Sorry.
John Atkinson
Editor, Stereophile
Response to John Atkins' Review of Compaxx mp3(May 24, 2006)
-----------
Part 1: Basic approach & issues - Part 2: Tech items
Part 1) John - My sincere thanks for your feedback. You've raised some very potent and pertinent issues. Hopefully your comments and my response will stimulate rethinking here, and others will join us in this dialogue. This is the goal of the Stereophile forum as I understand it. Beyond the basic exchange of knowledge, by interacting here we also gain some understanding and hopefully, a pinch of wisdom.
Please don't apologize for declining to endorse Compaxx on the basis of your expertise and authority as a stereo/electronics expert. I am foremost a musician, and only secondly a research scientist in audio physics. (My background is interdisciplinary, so I tend to think out of the box a bit.) I am by NO means an electronics engineer or expert, so my approach, tools, and goals are most probably very different than yours. For me, it begins and ends with the music itself. If the listening experience is pleasurable, people will buy it. Music that sounds better and gives more pleasure, sells better. The goal of Compaxx-mp3 is to make portable music content itself as pleasureable as possible, not to conform to engineering specs (which are important, and I'll address in Part 2 of this).
IMPORTANT: I find it fascinating that your comments regarding your listening experience of Compaxx as opposed to your lab analysis vary so extremely. As a music listener you state that you find Compaxx "vivid, more detailed, and more reverberant." (vibrant?) These are VERY positive comments, which you then discard and contravene because your analysis specs don't conform to your listening experience. You MUST decide in this debate on Compaxx which side of the fence to come down on, John. What comes first to you, a pleasurable listening experience, or engineering specs? For me personally as well as professionally as a producer, pleasure in listening will ALWAYS come first. Likewise with consumers.
Stereos are a tool for enjoying music, not a goal in themselves (status- and ego-challenged minorities notwithstanding). I've followed the debate elsewhere in Stereophile regarding musicphiles vs. audiophiles. It's a subjective argument, but in the marketplace it's much more objective. The music industry in the US generated over $11B in hard copy sales last year (RIAA claims $14B), and a similar amount in Europe. The stereo electronics market doesn't even come close. In the new world that's emerging, I envision the music and electronics worlds converging and becoming one, so my approach is a bit more holistic. The pleasure-giving qualities of the music content itself is the single most important market factor (in addition to portability, multiplatform useage, value pricing, etc.). In the music business there are 10 Commandments. They are: 1. Content is King. 2. Content is King. 3. Content is King, et al thru 10. By design, Compaxx is applied to content, not playback devices.
Summary Part 1: Compaxx is designed primarily for music lovers, not electronics or stereo afficianados. Which is REALLY more important to you personally, John?
Part 2 (techie stuff) Ok, I'll get the following out the way up front: I do NOT use volume or multiband compression, nor do I apply reverb. No way, no how, not EVER. Please rethink, and try again.
Issues you've raised...
a) Peak Volume Cutoffs: I'm not surprised that, after decoding to wav despite my warning, you produced a few (4) minor peak cutoffs in the Vivaldi sample. When stripping the mp3 envelope during decoding you've removed critical cuing information, resulting in a 2.2% increase in volume level with resulting cutoffs. Therefore, when I produce Compaxx in PCM format for CD application, I do so at a 2.2% lower peak volume than when producing Compaxx mp3. (On the Compaxx Benchmark CDROM I present both versions. When measured WITHIN the mp3 envelope, the peak volume is idential to the CD version.) Further, the original Deutche Grammophon master had several peak glitches. These were plastered over in postmastering by depressing the volume for these sections, as well as the filtered removal of certain fast-breaking sounds in upper bandwidths. Sometimes in restoring data, glitches and defects may reappear.
b) RMS: Yes, there are significant differences. Many times Compaxx restoration leads to higher relative amplitude due to the significant increase in playable data. (In Compaxx remasters of tracks that have had volume compression applied such as Enya's "Book of Days", the RMS will be lower.)
> 3 things about the RMS issue:
1. Can you raise the RMS of the CD original to the level in Compaxx (without using volume compression) without creating massive peak cutoffs and other ugly distortions? I invite anyone here to try it.
2. Have you compared the dynamic range between the files? Try it. Compaxx has a wider range, hence no volume compression. The answer is a wider bandwidth flow of MORE data, not pumping up the RMS. (That's ITunes' special trick, not mine. EAX does the same, but throws in reverb as well.)
3. VITAL: The zero crossings tell the tale. Zero crossings are strictly the result of the amount and type of data read and played. Compaxx remasters have more zero crossings (by 10-30%) than the originals even when compressed, because they contain MORE data. This is particularly so for fast-breaking sounds in the upper ranges. In graphic spectrograms, you can actually see the additional data presence, and in playback detail and articulation it's easily heard as well (you'll note the appearance of instruments and sounds that don't 'exist' in the original). Zero crossings, nuff said.
BOTTOM LINE: The REAL question remains: Does Compaxx SOUND better? As you yourself said about listening to Compaxx: "(Compaxx) is vivid, more detailed, and more reverberant (sic)."
Trust your ears, John. The ball's in your court!
With utmost respect,
Tuvia Wertheim, Compaxx Producer & Stationmaster
P.S. Brightness: It's easy enough to trim this, but an in-depth survey of 10,000 Compaxx subscription listeners shows that 89% prefer it this way. The consumer is always right by me.
I am just reporting what I heard and what I measured.
Exactly the same increase in volume was noted when I played the MP3 files with Winamp and Windows Media Player. Do these players also remove the "MP3 envelope"? I don't think so.
With all due respect, this is nonsense.
The post of mine to which you responded specifically compared the dynmaic range. Contrary to your assertion, the Compaxx version has less dynamic range.
I am beginning to suspect you do not know what the words you use mean. If you get more zero crossings, it is because you are increasing the signal's high-frequency content.
I do trust my ears. Nothing I heard is at odds with what I subsequently measured. You talk about improving the performance of MP3 at the low end of the dynamic range window, yet your process also introduces gross audible changes at the upper end of that window, as far as I can tell. It's a bit like telling someone to try to discern subtle improvements in the color reproduction of a photograph while you shine a spotlight in their eyes.
John Atkinson
Editor, Stereophile
Late here, so I'm going to sleep on this one before responding. It appears I've unintentionally pushed some buttons, for which I apologize.
Tuvia
P.S. mp3 is here to stay for a long time. I think that discussion here of the best way to produce it is beneficial to everyone. I also think that feedback from other forum members might add substantially and round things out a bit. I've received over a dozen email responses to the posts here from active forum members who are moving to iPod or similar, and encourage them to post in here as well.
I've also noticed that many forum members have logged on to my 192 classical listening libraries (URL in first post), and are spending a great deal of time there. Thank you for listening; let's hear your feedback.
And thank you, John, for making this forum available to us.
Well, this is getting personal and acrimonious, so it's time to wrap a few items up, and move on...
I never intended to get in deep here on numbers, graphs, and scientific debate. As I said, it's about the music, not the printouts.
From John's list of tech items he's taken issue with, I'll address only 2 for now. For the record, I'd like to note once again that I am NOT an electronics or stereo expert, and don't pretend to be.
Zero Crossings:
John: "I am beginning to suspect you do not know what the words you use mean. If you get more zero crossings, it is because you are increasing the signal's high-frequency content."
Perhaps you're right, and I don't have a clue. However... While in the past it was commonly-accepted wisdom that Zero-Crossings related only to high frequencies, even Rice & Logan suspected that zc's may indicate sympathetic activity throughout the entire spectral range and be a valuable indicator of total bandwidth activity. In the ongoing development of the Gaussian filters, this has been taken into account, even though this approach is still in it's first stages in terms of both theory and application. B. Kedem has published some excellent papers on this for those who wish to explore further.
Again, what really matters most? What we hear and enjoy, or what the numbers say? (Perhaps instead of spending time listening first, we should just run out and buy the stereo system with the best specs. It would be ever so much faster and intellectually satisfying.)
WinAmp: I am intimately familiar with WinAmp, and was one of Justin Frankel's first financial backers for it back in 1997. WinAmp is exceptionally gentle and forgiving in its handling of mp3, which is why it's still by far the most popular mp3 software player out there. HOWEVER, Compaxx was primarily designed for hardware playback, and this is where its full potential really shines. While people do use their PCs for listening and preview, the overwhelming trend is towards portable (and perhaps the emerging HD console) players for extended listening. This is why I recommended analog hardware playback for your review. I could go on, but won't.
I see that attempting to discuss here a better application of today's popular portable formats is not only highly provocative, but quite possibly an affront. Stereophile IS about hirez top-end stereo, not 'lowrez' media, no matter it's quality or popularity. I do get it. My mistake, silly me.
A forum member emailed me: "I'd be totally embarassed to admit on Stereophile that any form of mp3 is acceptable. I'm also ashamed to say out loud that I'm beginning to spend more time listening to my iPod than my obscenely expensive stereo. If I could get my compressed audio to sound like Compaxx, I'd just sell it all and go with iPod." Perhaps this is the answer to my questions of yesterday to forum members.
I suspect that stereo orthodoxy as it appears here will shortly become a thing of the past, a select minority aside. For nearly 40 years our cities and malls were all aglitter with expensive camera emporiums. How many remain today in the digital age? How long will the top end stereo emporiums and magazines survive?
I'll leave it to others here to debate this if they wish. For my part, I'm taking an extended breather. If other members post here, this will stay alive. If not, this topic will expire in 30 days and be relegated to the boneyard. It's up to you folks now.
From my first post here through this, all has been said with respect, and in the sole interest of creating enjoyable and fruitful discussion.
Regards until next,
Tuvia
I am sorry you feel that way. I wasn't aware that I was getting either personal or acrimonious. I was describing what I heard, what I measured, and the fact that my experiences were at odds with what you claiming.
But it's also about the claims you were making. Are you suggesting that it is not appropriate for audiophiles to question those claims?
Not at all. I have said that your treatment might well make MP3s sound better, particularly on compromised-quality desktop systems and particularly to naive listeners. The reduction in dynamic range, the increases in loudness, bass, and treble, what appears to be an increase in reverberation, and what appears on headphones to be a widening of the stereo image, are not unpleasant. I am merely pointing out that there is no need to make claims about "data mining" and reconstituting missing information to achieve these audible changes.
Regarding your comments about the Compaxx process needing to be auditioned on "hardware" players, I have since listened to your comparison files on my iPod and on iTunes running on my server. My feelings remain unchanged: Compaxx appears to use some sort of multiband compression -- the resultant "breathing" on the bass on the Joni Mitchell track is very audible -- perhaps with other effects involved. I am fully prepared to accept that the process is not trivial, and probably needs to be optimized for each recording. But I don't believe it is equivalent to some sort of "missing data" reconstruction as you claim.
Get what? If you were achieving what you claimed, I'd be the first to applaud. As it stands, you merely seem to be using this forum to promote your commercial activities using claims that I don't believe are supported by the comparison files you offer on your website.
Perhaps I am weary -- it's been a long day -- but I am getting tired of posters who make claims, react badly when I examine those claims and take the time to post my thoughts, then declare, absent any actual debate, that they are going to pick up their ball and go home. If you wish to contribute to this forum, Tuvia, then you should understand that you won't be treated any different from anyone else. Ask any of my writers. Ask Jason Victor Serinus, who underwent some pretty tough questioning on this forum about something he had recently written in the magazine. :-)
John Atkinson
Editor, Stereophile
As one whom you say led you to begin posting on this forum, and one who enjoyed your early interchanges with John Atkinson, I am both surprized and disappointed at your petulant "take my ball and go home" posting. Guess I had you figured wrong. Too bad you've decided to leave here, you might have learned something of value. Most of the rest of us have.
Hi, Tuvia --
I, too, am disappointed in seeing this exchange end. As a techno-idiot when it comes to the so-called "wave of the future," MP-3, I was actually learning something. Since my expertise lies in the area of rhetoric, I ought to note something. There is no acrimony in JA's commentary. He is always balanced and fair...and patient. His style is direct and matter-of-fact, as befits someone constantly bombarded with technical questions and requests to "just have a quick listen to my new technique." He must receive thousands of these per year. As I read it, JA invested a lot of time checking out what you asked him to check out. This, with the imminence of the HE show next week and thus a bazillion details pressing hard. I suspect you interepreted some of his directness as a lack of courtesy. This would be wrong. So, as Presidential Candidate Wong would say, don't go away mad. Just because JA disputes your claims doesn't mean he pans your results. He merely deems them a qualified success in a less universal context than you had hoped. I think his positive comments speak for themselves. We are all, in a sense, perfectionists when it comes to claims, and honesty is always appropriate when perfection is not achieved. Cheers, and best of luck in your endeavors. Clifton
I posted a spectral analysis of the original MP3 of the 4 Seasons extract and the Compaxx version in the Stereophile photo galleries. The dynamic range compression and equalization changes are clearly evident.
BTW, I think the Compaxx remastering has been done with care and I am not surprised if many listeners prefer them. I don't believe the changes are trivially easy to do. It's just that Compaxx doesn't need to invoke mysterious and proprietary processing to explain what they have done.
John Atkinson
Editor, Stereophile [image]http://forum.stereophile.com/photopost/showphoto.php/photo/558/password//sort/1/cat/500[/image]
What is interesting about the graphs is that the level shifts are not equal with the 65 hz shift at 15db and some others around 10db ( your comment of multiband comp), which is obviously a choice made in the compaxx process. My question is the levels as they are shown on the graph the actual playback levels for each sample, which seem recorded/transferred at lower than normal recorded levels, certainly way below 0db.
I am convinced that normalizating takes something away in the process, but some not into absolute total resolution retention , say for automobile playback, normalizing is adequate.
I also thought you to say that you found some "overs" which in digital can be a very bad thing depending upon their level. I read one recording engineer say that once you just turned on the last bit going higher did not matter. His argument would still leave 5+db of headroom or lost S/N, but really, at 90db of theoretical noise floor that seems to not be that important. More important not to go over I would think.
I look forward to really scrutinizing these files over the next week.
THE FFT analysis is showing the levels in each of the 8192 frequency "bins." If you sum these individual levels, they will add to give the overall peak level given in my earlier posting.
I don't think he's correct in his understanding. If he's referring to the Most Significant Bit (MSB), this is turned "on" once every cycle when the signal goes negative. If he's talking about the Least Significant Bit, the difference between the final time this is turned on is 1/65536, ie, very small. You've already used up almost all your headroom at that point.
Let us know what you think.
John Atkinson
Editor, Stereophile
Dear Stereophiles:
I apologize sincerely and without reservation if I've offended here, it's the furthest thing possible from my intent.
I also wish to thank John here for taking time from his extremely busy schedule to evaluate and comment. I can see how much you love and care deeply about your work. The world needs more people who truly care. I thank you in particular for your patience. I note that you've come halfway to meet me by expresing yourself yesterday in listening and music terms I can understand. This is greatly appreciated. Thank you for all of the above.
I haven't said that I'm leaving here, just taking a breather until others join in and expand the dialogue. As long as it remained strictly between John and myself it was becoming increasingly difficult, as well as defeating the purpose of a forum which is to generate discussion, not private chat. I'm thankful that others have joined in, as I believe this to be a worthy topic of general interest. So, thanks to all who joined in for helping expand. As long as this topic stays alive, I'm in. Please try to overlook my style, we musicians aren't easy people.
If there are specific issues I don't respond to, it's either because of my lack of professional knowledge on your terms, or shortness of time. (John has his trade fairs and clients, I have my own. It's summer music festival time once again again.) I'm not dodging here. I'm a music person; you folks are stereo experts. It's not easy to have a dialogue on this basis, but I believe it vital to try in a world where are our mutual spheres are becoming increasingly interlinked. (As Abe Lincoln put it succinctly: "Gentlemen, either we all hang together, or we'll all hang separately.")
Please note that I am NOT using Stereophile to promote myself in any way. I came here to invite people to listen, enjoy, and explore the possibilities of the emerging world of portable digital as I am attempting as a producer to enhance.
Believe me, not a single one of my listeners reads audiophile magazines of any kind, so there would be little point in promotion here. If I thought so, I'd pay and take out advertising. Your audience spends their money on hardware, mine spends no less than yours, but on music. And while my audience may not have state-of-the-art systems or be professional listeners by stereophile standards, they all spend a minumim of 6-8 hours daily listening to music and care very passionately about what they hear. All of them play instruments, so they know what's authentic in sound, unlike those who grow up on a strict diet of CDs. A significant number are professional musicians and artists. Their feedback is what drives the direction of my work. In summary on this, I haven't approached John to solicit a review, though I am extremely grateful for it and have learned much. I greatly respect your professionalism, and appreciate your time.
-----
On to issues (in the positive sense)....
Although primarily intended for music-philes, I do welcome audiophiles to listen to Compaxx, and feel I have a great deal to gain from their feedback. I've learned a great deal from this interchange itself, though it's hard for me to tune in properly to a jargon which isn't my native bailiwick.
Sidebar on the above: Tone is as important as content. I'm having a hard time not being provoked by what appears as backhanded dissing, hence my defensiveness. Example: "I have said that your treatment might well make MP3s sound better, particularly on compromised-quality desktop systems and particularly to naive listeners." Please try 'lowrez' and 'non-stereophile' instead of "compromised, naive". It gets the message across clearly, without offending.
> Data-recovery issues: I'm having a problem discussing this in here, and it's probably the result of my lacking your perspective and vocabulary. I've stated my case as clearly as I can for my listeners on my webpage, but here I find myself fumbling to express the concept in your terms, and apologize if you feel misled. 'Data-mining' was probably a poor choice in my attempt to fit my concept into your jargon, and my apology is rendered.
To the point - No, I do NOT recover non-extant or damaged data, nor do I claim such. What I do, and state very clearly on my webpage, is make RESIDENT but unheard (or relatively unheard) delicate sounds of specific types hearable (not an easy task given mp3's extremely restricted bandwidths). It's like being in a well-furnished dark room with only a candle; all you would see is a small circle of light and maybe a chair or two around you. Turn on the halogens (or spotlight as you call it), and you see everything. I'm not a magician conjuring up the furniture when you're not looking. (Nor am I a psychoacoustic wizard dropping LSD into your mix to make you see things that aren't really there.) I simply 'turn on the lights'. You may prefer that they be incandescant rather than halogen (the brightness issue), and that's a legitimate and highly worthy comment I take to heart. Compaxx is restoration in the musical sense, not in the way I now understand stereophiles and DSP folks use this term.
Case in point: Joni Mitchell. On the original CD recording the bass line appears to be a regular solid-body instrument. You now notice the bass 'breathing', and suspect it's some type of reverb or air techinique. If I recall properly from seeing this piece performed live, the instrument used was a hollow-body horizontal dreadnought hybrid bass by C.F. Martin. (I myself play a 1972 Martin D18s, and am a great fan of the Martin sound.) The Martin bass is an unusual instrument used by several folk artists at that time, and has a very rich and resonant sound.
Way back when, my start in music was as a folksinger back in the 60s-70s. Guild also had a similar bass back then, and I hear Taylor also makes these today on a custom basis. I've played with many of these artists, and seen, played, and know their axes well. It's lovely to be able to hear this instrument come back to life after turning on the lights. (note: I used to have the original LP of this, and the hybrid bass was more apparent there.) I'll check with Joni if I can and get back on this regarding the specific model. (btw, I hear that Joni's been fighting a hard battle with cancer. If you go to her webpage and drop her best wishes, I'm sure it would be appreciated.)
---------
Regarding mp3 as a worthwhile way of listening in terms of quality, here's how it was put to me in 'non-scientific' terms by a well-known rocker: "mp3 is sh*t, but there's good sh*t and bad sh*t. Compaxx is PRIMO sh*t!" (pardon the french, but you get the idea.)
All in, I find John's comments on what he's hearing very positive and enlightening. I have a great deal more ahead of me in getting Compaxx right; it's a work in progess.
Thank you one and all. Hang in please, and we'll enrich each other's lives. What else is music for?
Regards,
Tuvia
P.S. John notes the pleasant effect of space and 'wideness'. I hope you find it pleasing and natural. This is perhaps a topic for next time.
P.P.S. Cheapskate: I've been visitng these forums as a non-member for some time and have found a great deal of thought-provoking discussion and valuable info. It was your reference to the IPO's Vivaldi that drew me in, as I know the recording and it's background. Thank you for leading me here.
P.P.P.S. Jim T: I caught your comments. What you note is intentional, and by design.
Tuvia - While it might be disingenuous to say we're not gearheads, I think you may have a skewed view of audiophiles (at least the ones that post in this forum) when you say, "Your audience spends their money on hardware, mine spends no less than yours, but on music."
Most us here have spent more on hardware than the average Joe, but, it's precisely because we are music lovers and want to hear it in the highest fidelity. Sure, I might have 3 systems... the he-man system in a mostly dedicated room, a headphone rig, and a bedroom setup... but, and it's a big one... I have thousands of LPs and thousands of CDs, as I suspect many other members here do as well. A lot more money is getting spent on music than gear.
I am sorry you perceive my comments as "backhanded dissing." I am offering my comments on what the Compaxx technology appears to do. I have posted spectral analyses of the "before" and "after" versions of the Vivaldi comparison MP3 files you posted on your website in my personal page in Stereophile's website photo gallery. This morning I posted the original waveforms, the original MP3 and the Compaxx MP3:
http://forum.stereophile.com/photopost/showgallery.php/cat/500/ppuser/31
I am not intending to offend anyone with my usage. You are incorrect to refer to the iriginals as "low-rez" as they have no less information that the Compaxx version. And by "naive" I mean someone with no experience of critical listening, with limited awareness of how trivial changes to the signal can have large effects on preference.
If you look, for example, at the waveform plots I posted, you will note that the Compaxx version is 12dB louder than the original. This _ in itself_ will lead a naive listener to prefer it in A/B comparisons, but of course, the preference would be eliminated by playing the original MP3 the same 12dB louder. Add the changes in treble and bass content introduced by the Compaxx process to this loudness difference, and I am would be surprised if _any_ naive listeners wouldn't say they preferred the Compaxx.
As I said, such changes are "trivial," in that they can easily compensated for in a fair comparison. But what I object to on your website is the fact that the comparisons you offer are _not_ fair. If I _did_ want to diss Compaxx, I might even suggest that you had no intention of making these comparisons fair, that you _wanted_ inexperienced listeners to come away with the impression that the Compaxx files sounded better and that they did so, ipso facto supported your claims about restoring missing information.
But I won't say that. You appear to genuinely believe that what you are doing is a step forward in sound reproduction and I respect that belief. I will just leave it that my measurements and my listening both indicate that what Compaxx appears to be doing may lead many to prefer the sound, but that does not in itself confirm that your technology does what you claim it is doing.
I look forward to your comments on the differences that I have noted in the captions to these images.
John Atkinson
Editor, Stereophile
Wonder how the Compaxx technology compares to the EQ the Burwen Bobcat applies to MP3?
Great to hear it; I was starting to wonder. Whew, people who don't listen to music make me nervous. What are your favorite genres, pieces, etc. When you take a disc to a stereo store to check out new gear, what's your 'benchmark'?
You've made my day Wong, and restored my faith in stereophiles! Keep on posting.
Could you tell us more about this, such as where/how you've heard it, what do you like in the effect, etc.? Compaxx isn't the only gig in town when it comes to working on nextgen quality mp3. There are quite a few groups and individuals burning the oil to bring the world's portable digital music up to a worthier level. What's ESPECIALLY interesting and different at this time, is the work being done on the music itself (post-process) prior to encoding on extant codecs, as opposed to reinventing/inventing new and better codecs.
I'll read, think, and do my honest best on it, John. Believe it or not, I really do look forward to yours as well. Your comments warrant some serious thought and time, so I'll only get to this Sunday or Monday. Ego isn't a factor for me, so if through your feedback I learn that things need to be changed or made better, I am more than willing to go with it. I'm not in Compaxx for the fame (I'm a pretty private person) or the money (this has been non-profit for 10 years already; my income derives elsewhere.) This is a search for the best possible way to render music in today's preferred portable formats, so my mind is open to what I learn here. I do use Compaxx as my foremost tool in publishing and broadcasting (my listeners and clients love the sound), but it's not something I sell. Just wanted to say that and dispel the notion that there might be an agenda here. My only agenda is to share and learn with you folks.
> In addition to responding to John, I'll be posting some new files to my site with url's here just for Stereophile. These will include both original wavs of CD as well Compaxx redos, in full length versions. 60-sec doesn't work very well and really can be skewed in terms of relative dynamics to a whole piece. On my public site I'm constrained to limit pop samples to only 59 secs so as to conform to copyright issues, and only use mp3 vs mp3 there so as to conserve on general listeners' bandwidth. You folks are definitely NOT general listeners. (LOL)
Please note that the link to my classical listening library (http://www.compaxx.org/php/00pvt.php <-copy & paste to new browser externally) was created only for people here and is not for the public at large. I notice that since last Wednesday some Stereophiles have been listening in at length (about 5 gB's daily) and thank you. As long as I can handle the bandwidth and the visits continue, I'll leave it open for Stereophile.
Keep Posting!
Some topics I'm thinking of including in future posts here for general chewing in relation to our shared hobby/interest:
> Dolby audio cassette hiss removal - If it were now 1972, and we were there with Stereophile, how would we react if Dolby joined this forum and posted about it? ("Audio cassettes for stereophiles???? hahahahah" or "Hm, can/should this be really be done?")
> How do we hear? The relationship between physiological & neurological evocative cognizance and audio signal modulation. (This is where I started with Compaxx while on a project in NC, 1995. Duke Universtiy is a world pioneer in hearing impairmentment research, and has some excellent papers bearing on this topic as well as the nature of hearing itself. Of course I wondered at the time how this could be used in music, and here I am with Compaxx today.)
P.S. To Jim & John: Jim caught something at the end of his last post which derives from the above (frequency pattern modulation) from Duke, and is by design. Without going into details at this time (long, and probably esoteric), it's one of the results of 'phasing', which I use significantly in Compaxx. This is perhaps why John perceived reverb or multiband compression. I do use equalization of various types, but again without compression. Also, please note that Compaxx runs at 48k, not 44.1k in mp3 (several good reasons for this), so if you're using a software decoder please don't resample (dithers can sound awful).
Please tell me how you'd recommend doing this. I work on Sound Forge, and if there's anything I'm capable of doing to obviate the need for compensation to make comparisions more fair, I certainly will.
Just for general info: Compaxx is a multi-layer offline process. There are 35 steps (process layers) applied to the entire file one at a time on top of one another prior to encoding. The effect is cumulative. STF process time on a Pentium IV with 1gB RAM is 2 hours processing per 1 hour of source audio (2:1). (Just in case you were wondering.)
Keep on Posting!
Tuvia
BURWEN BOBCAT is a plug-in for the Windows Media Player 10.it plays only through a USB connected Burwen Bobcat
Digital to Analog Converter made by Daniel Hertz Advanced Audio designs
http://www.burwenaudio.com/BURWEN_BOBCAT.html
I've had a look at Burwen, and given it a trial spin. There are several excellent DSP plugins for both WinAmp and WM. Many are also in hardware form (such as DFX and EAX, which come in both).
My philosophy in developing Compaxx is to embed the DSP within the music itself, rather than creating an external plugin.
Thanks for posting about Burwen. I'm sure many here will find it effective and useful.
Tuvia
Amongst others, Jim Tavegia has been generously doing some serious homework on Compaxx and emailing his initial comments. I've asked his permission to post excerpts from his emails here, as I think it will benefit everybody. I believe that Jim's responses together with JA's analysis and questions will help paint a more comprehensive picture. Also, Jim speaks excellent 'stereo talk' (which I don't) which helps clarify in this forum's jargon. - Tuvia
May 27, About Compaxx 56k and 32k files (Jim will only be commenting on the 192k after working on hard-copy later this week.)
"Tommaso Albinoni: 56K file. Very nice. Full sound with decent power at the low end. Certainly the quality of a good NPR classical broadcast on a good FM stereo tuner. Just because it sounds like analogue is not a bad thing to me. I own three turntables. I could listen to this quality all the time. I hear nothing artificial in this recording or stream."
Beethoven/Emperor: 56K file. Nice sound, but a slight bit too much reverb on the piano. I do not seem to notice the reverb on the rest of the orchestra. The extra reverb make the recording slightly bright. This is one of my favorite pieces. When it comes to the piano I really am a purist. Longer into the piece I do hear the reverb on the orchestra sound tails."
"Chopin Concerto 1: 56K file. The sound when playing was very good, natural sounding. I have lps that sound much worse than this stream. A full bodied, smooth recording and sound stream."
"Mussorgsky/Pictures: 56K. Nice sound appears as in a large hall, long decay lines. Very natural to me. Good low end extension."
"Tommaso Albinoni: 32K Nice sound, with some minor loss of smoothness, but still good low end and violin character."
"Beethoven/Emperor: 32K file. Little less resolution, but still to much reverb on the piano to my liking. It does not sound natural. Sound is still well balanced freq wise. This is one of those times I wish I had a pair of AKG 414 BLIIs and a DW Fern 2 channel vacuum tube mic pre and got to do the recording."
"I found the wave forms that JA put up interesting. Plenty of dynamic range left in for sure."
Given that FM is broadcast at 96k/44k and the Compaxx 'FM Quality' is only 56k/22k, I find Jim's comments ("Certainly the quality of a good NPR classical broadcast on a good FM stereo tuner...I could listen to this quality all the time.") extremely helpful. - Tuvia
May 28: "I am listening to your streams through my TV Room Pioneer Elite 5.1 HT system. I have listened in stereo and through all the Dolby Digital and DSP settings for enhanced sound. I have enjoyed all the music played in every format. The Beethoven Emperor still has too much reverb for me, but the rest are very good."
Thanks, Jim, extremely helpful. Please post your comments about the Compaxx 192k when you're ready.
Other Stereophiles are invited to join in and respond to the above!
Tuvia
P.S. Best wishes to all attending the SHE exposition this weekend. Have a blast, and keep your eyes open for what's new in mp3 for Stereophiles.
I want to clarify one statement: When I said all formats, I meant Tuvia's 56 and 32K formats, not all of the DSP formats in the Pioneer. I find all too many of these to be nothing less than awful manipulations of the sound. That is the price we pay for loving 2 channel.
In my Pioneer Elite Receiver I like Theater 2 in DSP. Some of the Dolby Digital settings not as extreme considering the appropriate source material.
I have some concert videos I enjoy in DTS as well as Dolby Digital and some I just listen to in PCM. I have been listening mostly to the Compaxx files in 2 channel.
The Compaxx(tm) mp3 enhancer: (http://www.compaxx.org/)
In comparison, the Creative X-Fi Crystalizer (http://www.creative.com/products/product.asp?category=1&subcategory=208&product=14065 http://www.soundblaster.com/products/x-fi/technology/music/):
What the Crystalizer actually is as reported by Digit-Life: (http://www.digit-life.com/articles2/multimedia/creative-x-fi-part2.html)
A (glorified) multiband compressor.
Axon,
Thanks for your reply. As I noted yesterday, there are many external DSP 'enhancers' out there today, both software plugins and hardware. Some are very good, others less so. As mp3 is temendously popular today, many of the people who use these are compressed music collectors rather than CD/SACD/DVD listeners. There is a tremendous interest in improving playback for compressed formats in mp3 in its current form, as it's so popular.
Most of the enhancers I've tested seem to rely on 2 basic methods, reverb and EQ manipulation (either bass boost or multiband compression). These are applied either together or separately. Some folks swear by these devices, others swear at them, and consider them gimmicks. I have to state again here that I don't use reverb in Compaxx, nor do I use EQ multiband compression. I don't even own the tools for these.
What I've done in Compaxx is to embed DSP within the music files themselves so as to eliminate the need for external devices. I've been even a bit more audacious in trying to create in Compaxx a multilevel process that works equally well for all types of acoustic recordings and all genres. (Since a large part Compaxx works on harmonic resonance, the process does very little for midi and electronic pieces.)
You'll note that all external enhancers use a variety of presets for different genres (classical, rock, etc.) If the process works properly, this shouldn't be necessary. So, whether you may like the achieved sound of Compaxx or not, Compaxx works equivalently accross the board.
I'm a music producer and broadcaster, not a hardware or software developer. As such, I believe that music that sounds better will sell better. I am also solidly convinced that the market opportunities and future belong to compressed formats for the wider market, and that mp3 is here to stay. The only serious limit to widespread commercial sales of mp3 is its inherent poor quality. This is what Compaxx was designed to overcome.
I believe in music.
For those serious about DSP plugins for high-end rendering (especially analog fans) you may wish to try iZotope Ozone 3.
so.....
I have been listening to the sample tracks - I have poked around the library a little, but I will give it more of a listen.
I am listening with Etymotic ER-4S from my IBM T40 laptop.
My take on this interesting post and technology. Some may like these tracks, some may not. I find myself in the latter group.
With the high resolution of the ear canal phones, I am getting some strange effects that I am not liking in the end. My first reaction was that it was louder and that there may be some cross cancelation going on (kind of like the circuitry in Headroom amps). This would give you a wider soundstage presentation, kind of like the old Polk Audio SDA systems. But I have no problems with any of these.
I am having two effects (artifacts):
1: WA WA Effect.
I could not come up with a better term. There are times when there is a center image (things in the middle (voice, piano, guitar)), I find am getting the image slightly moving back and forth to the left and right, like someone swaying back and forth (but fast). This is giving me a slight wa wa effect, and not a rock solid image of the instrument or person being in one place. I think that this is also what I am getting in the large orchestra hits that is making them seem muddy to me.
2: Strange instrument attack
The attack of almost all instruments has changed so that they do not sound natural to me. Let me see if I can describe the effect that I am getting. On the initial attack of the instrument (drum, strings, voice), there seems to me almost a slight delay (or decay). If I were to draw a simple graph of the standard playback you would get an attack _/^\_, but I am hearing some articulations, it would start from the bottom give a slight waver and then peak _w^\_.
Now, I am not saying that this is what is happening in the recordings, I am just trying to describe what I am experiencing.
I think that you are doing some interesting work, but I find it more distracting than engaging. I kept thinking from the beginning that something seemed wrong. You might not find your audience here, but there will be people that like what you are doing (maybe Head-fi).
Dear Stereophiles,
It's been a holiday week here, so my apologies for the lag in posting. On Sunday I'll write and post a summary of this chain to date.
In addition to the critiques and posts here I've received over 4 dozen emails from this forum with feedback and questions. I'll do my best to respond and summarize in operative terms what I've learned here.
Thanks to everyone who's participated thus far,
Tuvia
After nearly 2 months have passed it seemed like a good time to report a follow-up on the Compaxx process. What one must keep in mind is two fold: as a potential business opportunity and as a listening experience.
With all or most who grace this forum wanting high rez formats to live on and be supported by the music industry Compaxx is not what you may desire as many consider mp3s the black hole of audio. The masses have seen it all differently with their votes going to IPod, ITunes and the like. Portablity and convenience reign.
My continued two months worth of discussions with Tuvia have centered around the process of Compaxx: how much and of what flavor. This past week Tuvia sent me some new files with just A,B, C designations on them to keep it sort of a "double-blind" technique.
What had become clearer to him and to me was that possibly 2 processes might be needed; one for mostly acoustic performance recordings and one that is more pop music oriented with midi instruments and heavy compression typical for radio airplay. That proved to be the case in this latest round of testing. Less Compaxx process on acoustic performances and the standard process for most everything else.
As Tuvia proceeds on his journey of marketing Compaxx as an enhancement tool for musical enjoyment we must keep in mind that this is driven by a whole new set of parameters of acceptance by the music buying public. It can be easy to starve to death catering to audiophiles.
The idea of having good quality performance of all 9 Beethoven Symphonies on one CD I do find intriguing. 8+ hours of music on one cd that would make selecting music for a road tip much easier; 4 hours of driving each way and not listening to the same song twice. The idea is to find a way to make the music fun and enjoyable enough to listen to that someone would buy a Compaxx disc with 8+ hours of music at "some price" acceptable to all parties. It may also become the gift that YOU give that keeps on giving to some member of your family, even though it is not your "cup of tea".
I have found lower bit rate files listenable when I have compared them to some of my older LPs that I cherish. It is hard to not love totally black backgrounds with no clicks and pops and other surface noises. Given the immediate option of DSD or MP3, DSD wins hands down, but there is no way to get 8 hours of DSD on one disc...yet. Sony's lack of support for their own format is troublesome enough for audiophiles.
JA has touched on the new problems with Blue Ray and why he is not hopped up about it. It is the typical music industry look (listen), but don't touch. We can only hope that DSD and 24/96 can survive and that at least the "Masters" that are now being created can be put into what ever new formats that come about that will preserve more of the original performance than we have had in the past 60 years. We would all be rejoicing if Kind Of Blue Masters were in DSD. They were pretty magical as they were on mylar. Maybe taking up less space will keep more of them around as well.
The idea is that if the Compaxx process can and does make more music immediately available in a convenient format that they find pleasing is enough for me to see where it goes. I can see endless possibilities given the right opportunity. Since most of us listen on something less than B and W 802s it is clear that there is room for lesser formats as we all are living with less. It is fun having the high rez formats pump us up as high as we can go though.
Great post Jim! Tuvia is very fortunate to have found one among us with the open mind and willingness to explore his endeavors. Great bleepin post!
RG
Rgibran,
Thanks for the note.
I have found all of this interesting and educational. I had the older version of the Fraunhofer IIS codec in my Cool Edit program and did some research with my waveform editor and found a huge level reduction at 13K of almost 10db. THIS IS A VERY BAD THING. A low pass filter no doubt, unintended for sure?
The LAME codec in my NCH Swiftsound Switch program is ruler flat. This wetted my appetite. It is not DSD, but the files I fooled with were certainly listenable in my car and for casual listening. I have an older Solti/CSO Beethoven Emperor London LP that as I listen today is so worn out it is time to retire it. An mp3 at 192 would beat it. It was good that I bought about 10 years ago a new set of 1-5 that was still sealed. Hey, You never know!
With my 11 year old son, Nick, most likely wanting a PlaySation 3 when they arrive on the shelves I will find out what Blue Ray has to offer to a degree. I am guessing that will be the first Blue Ray device to make it into my home.
Tuvia and I are going to talk some more this coming week. We talked today about some unrelated issues of music sources. My crystal ball is so messed up I have no idea what the future holds. I do know that with all the marketing troubles besetting the audio industry at the mid-fi level they need the next "IPod" to show up quickly, what ever that is. I do not think that the IPoop: Ipod docking station with a built in toilet paper dispenser is really IT! At least I hope not!!!!!