First of all, I must point out that I am not the best person in answering the most technical questions. I understand the priciples of these issues quite well, but there are others that are much more compentent in helping you out with the tech stuff. Some old mates should come to visit this forum (from my old one) and they have the knowledge you seek. >I forget my experiences with BladeEnc, but >I don't recall any major dissapointments >offhand. Usually specific test samples are used to discover the weaknesses of codecs. People have found Blade failing in these cases. Blade is very much based on the original ISO source code. LAME originates from the ISO code as well, but has been optimized so much that the original code was 100% replaced quite some time ago. When comparing LAME and FhG, at least few things stand out. FhG is much more aggressive above 16khz, so it's more optimized for lower bitrates. At 128kbps FhG actually still can challenge LAME in terms of quality. The JointStereo implementation of LAME is best of the industry as is the VBR also. You should definitely update to the latest LAME 3.90. I think EAC+LAME.EXE combination is the best at the moment and once you get it configured, you'll find that using it is very pleasing - quick and simple. Commercial MP3 sites propably choose 128kbps in order to save space / bandwith. The royalty free OGG Vorbis codec handles lower bitrates better than any MP3 codec, so hopefully commercial sites adapt to that. I have never really used the Audiograbber so I really don't know about it's FhG implementations. Switching to EAC+LAME combination will give you SECURED ripping quality along with higher quality of audio compression, and both softwares are free so...
It seems that Blade is even worse than Xing codecs at 128kbps, or maybe at the same level with Xing old. Unfortunately some of the claims by the Blade developer, Tord Jansson, are regarded even quite humourous among Lame and other developers, especially his claims about "tonal purity". From Blade site http://bladeenc.mp3.no/skeleton/quality.html "The other side of the coin is that BladeEnc has an extreme tonal purity. While all (?) other mp3 encoders seems to distort sound in the frequency domain, BladeEnc keeps the tones sharp and precise." One should not believe any faery tales like this, and also be very suspicious what is said at bladenc site. The fact that Bladenc's pre-echo handling is totally messed up, doesn't make it more "tonally pure" even at 256kbps by any means. Bladenc is nothing more than only very slightly improved version of chaotic (quality wise) ISO-mp3 example code. PS. a small obvious correction to your message, cd-rw.org. The latest (stable) Lame is 3.91.
The Blade "tonal purity" myth is also propagated by such sites as Chris Johnson's at: http://www.airwindows.com/encoders/index.html Chris suggests that Blade's supposed strength in tonal purity at high bitrates is the advantage gained by sacrificing transient definition. Chris also propagates the mistaken idea that spectral graphs can be used to evaluate codec quality. In fact, graphs are exclusively used to support his contention that Blade has better tonal purity than Lame or Fraunhofer at high bitrates. And this graph nonsense comes from someone who actually makes his living by listening. In my opinion, even at high bitrates, Blade audibly degrades the sound of the music. To hear how bad Blade sounds, try encoding 41_30sec.wav with it at various bitrates, starting from 128 kbit/s, where Blade is just awful. I was able to reliably distinguish (ABX) Blade from the original up to bitrates of 256 kbit/s. 41_30sec can be found on my samples site at: http://ff123.net/samples.html ff123
I used Blade up until one day I was fumbling around in CDex and saw Lame at first I didn't think much about because of all the "hype" on the BladeEnc website. Anyways I tried Lame and wow what a difference the quality was so much better. Of course that's old news these days, and to me so is Lame since I've embrassed Ogg Vorbis.
andavari: (You might not still be around, but anyway....) I was just looking through some of these older posts, and I see that you have either "embarrased" (or) "embraced" Ogg Vorbis. I assume 'embraced', because I too am just *floored* by the quality of this compression technology! I've just spent an entire day ripping waves, and comparing their .ogg equivalents to (.mp3s mostly), and I just can't get over how *awesome* these .ogg files sound in general with any quality setting from -q 5 and up. I've been wedged in between a pair of Sony headphones for the last 3 hours now, doing some critical listening. At quality 10, my ears can't discern the original wav from the compressed file, but this is mega-overkill, because to tell the truth, I can't tell the two apart with settings substancially lower than that either. What an *incredible* achievement this format has become. I've used both the simple GUI encoder (drag & drop) and the command-line encoder, preferring the latter because of the more accurate settings I can specify. I wasn't really going to get into .ogg files, but then I noticed that the latest version of Winamp (version 2.80a, May 01/02) was the first one to offer ogg vorbis support, and so I couldn't resist. Not only that, but Winamp provides fool-proof tagging which is easier and more flexible than .mp3 ID3 tagging. I don't know if this file format will ever overtake mp3's current hold or popularity, but it's a most worthy competitor. More's the pity if it doesn't catch on, because I'd love to see some sort of hardware support (like, maybe, being able to play .ogg data discs in my standalone dvd player, just as it now handles .mp3 data discs). What are your experiences with .ogg? What quality-level do you usually use for high-quality (archiving purposes) cd-music tracks? Best Regards, -- KlingonAgent --
Klingon, If you think OGG is great, then have you tried MuspePack. OGG is still unoptimized and it has obvious flaws with some test samples, even though ut generally is above MP3. But MusePack is another story - MusePack is simply amazing. http://cd-rw.org/software/audio_software/audio_encoders/mp+_encoder.cfm
Hi! With your kind link, I have just d/l'd MusePack 1.04, cd-rw.org. (What is your name, friend? I have very high regard for the most intelligent folks at cd-rw.org, but the name is tricky to type-in all the time. Are you Bob, or John, or Mike or Steve?.....) You seem very enthusiastic about Muse, so I would certainly be amiss if I didn't try it, which I assure you I will as soon as time permits. We sure have come a long way from mp3, eh? I'm not overly worried about .ogg not being formerly optomized for speed - the time necessary to compress any given wave seems virtually inconsequencial after you have compressed a couple hundred DVDs - not joking about that - <gg>, but *quality* optomization, well, that's another whole story, and one, I expect, which will go on forever as new audio-compression algorithms and formats emerge. I would like to have and use what I think most people would like - a format that is stable, is going to be around for a while, is (preferably) open-source, carries no user-restrictions (DRM), sounds outrageously good, offers a decent compression ratio, is easy to use, and secondary to all of *that*, a format which, at least to some extent, offers the occasional free internet (music) download from time to time. In other words, a robust format that is recognized by content providers, at least a few, and used by enough people at home to help ensure its longetivity. Truly, ogg vorbis seems to encompass all of the above, and more. I did quite a bit of web-surfing yesterday, and found out that several sites are now, indeed, beginning to recognize .ogg for its strengths, and what my laborious testing yesterday confirmed - it's a FINE audio format. Just as an example, the folks at "angry coffee" - http://angrycoffee.com/tutorials/vorbis - are now recognizing the importance of .ogg, and I found at least a half dozen other sites, which I shamefully fogot to bookmark (I just checked my 'Favourites' folder - I forgot to log in a whole whack of *super* sites yesterday, who gave me some *super* sounding .ogg music files). Oh well, [shrug]. Your opening page to MusePack v.1.04 starts out, "MusePack is the highest quality lossy audio compression format currently available..." Well! (Ratz, I wish I had seen that statement earlier; might have saved some time) - so I'll certainly have to check it out. The official vorbis page, http://www.vorbis.com , seems to be down currently - as it was for most of yesterday, but I have several pages of printouts from them that make a *powerful* case for the format, not the least of which is ongoing development & support, and backwards-compatability with previously-made .ogg files. But perhaps I'll give Muse a try, and see if some of the rest of the Net is using it, (on a personal level) and if there are any sites offerning free promo tracks. Thanks for the tip, cd-rw.org! (Er, ... bill, frank, paul, ....(ummmm), what did you say your name was?) -- Michael, a Klingon of your acquaintance --
OK, really have to run......am preoccupied with earning a living, but briefly..... Ran MusePack 1.04 beta. Resulting file extension = ".mpc". What is this? I have done no web research on Muse or .mpc; was hoping you could save me some time here. Is mpc some subset/variation of some mpeg format? Neither Windows Media Player or WinAmp 2.80a recognized it. Do I need a codec? Do I need a standalone software player? I hope not. No help file included in musepack, but I generated one by redirecting built-in usage parameters to an ascii file thusly: mppenc --longhelp > usage.txt Resulting text file, which I printed out gave all the available settings, 99% of which I or no one else will ever probably need. Of the quality settings: * telephone * thumb * radio * standard * xtreme * insane * braindead ... I chose the highest (braindead) for a test. Quick specs: original music cd .wav file = 33,591,308 bytes ogg vorbis file at highest (q-10) = (about 9 megs) .mpc highest setting (braindead) = 6,058,804 bytes At highest settings, muse produces the smaller file. Have no way of determining its quality; I don't know what the file extension .mpc refers to. Enlightenment kindly requested, please. -- A busy klingon --
To play mpc files, you either have to decode them to wav using mppdec or play them using a Winamp plugin, which you can find at the same place you found the encoder. The recommended encoding profile is the default one (--standard). Musepack is based in part on the same ideas as mp2 (although the psychoacoustics have been greatly improved). The most noticeable improvement over mp3 is in reproduction of transients, which is nice and clear with mpc. For a listening test of this property, see this thread: http://www.hydrogenaudio.org/forums/showthread.php?s=&postid=19674#post19674 ff123
Ok, found http://musepack.org . Silly for me not to have guessed <gg>. Site still heavily in fledgling stages - only limited info, however have downloaded but not yet installed necessary winamp plug-in. Will do so when I get home after work tonight. Question: Do you really believe that musepack IS the ipso-facto mostest-bestest sounding audio file format in the universe today? [ ] - yes [ ] - of course [ ] - certainly [ ] - all of the above [ ] - none of the above [ ] - (what was the question again?) And also, why? Have you done direct comparisons? Is there a filesize/quality tradeoff or a filesize/quality advantage with this format? What are your feelings after using the format at different quality settings? What colour is blue? Why is water wet? What is/are your real name(s)? Have a nice day. - Mike -
Klingon, To your question: MusePack is definitely and absolutely the best. I personally am a bad listener, so my opinion is based on the opinions of experts, like mr. ff123 here. Also computerized sound file analysis support the fact that MusePack outperformes the competition. The -standard mode if said to be 99,9% flawless. Many people however go for -extreme, because it doesn't produce that bad bitrates. The WinAmp plugin you'll find at afterdawn/cd-rw.org - search "musepack". (Hm..I gotta add some MusePack front-end too..). Also, the Media Jukebox supports MPC directly. If you insist you can call me Lasse
Thank you, to all, for the speedy comback. As you can see, I still haven't got off my a## and gone to work yet. Being self-employed does have its perks. Well, cd-rw.org, "Lasse" it is then, whether that's your real name or not; the only reason I mention it is that *anyone* @ cd-rw.org can respond under the blanket name, and that can leave me scrambling as to who I am actually communicating with. Lasse> .....MusePack is definately and absolutely the best. Thank you for taking a firm stand on this, Lasse. I can now disagree with you. I put the proper dll into WinAmp's plugin folder, and was able to play back the .mpcs I created earlier. The tonal balance has shifted terribly with Muse, something I have not noticed with any other format, regardless of bit rate. I used the top three settings for Muse - xtreme, insane, and braindead. In each and every case, the highest frequencies were partially dulled, and the whole upper-midrange frequency band has been muted as well. And not just subtly, either. Oh, the sound is clear enough, but the highs have dropped significantly and the whole soundstage has been shifted. This difference is immediately apparent to me - it just kinda jumps right out and grabs you by the boo-boo, y'know? This is very very bad. There would be no point is trying the 'standard' setting because if Muse can't keep the frequency balance straight at it's three best settings, dropping the bitrate even further isn't going to solve anything. And in truth, it's much better to use the highest settings in these tests because one will more clearly be able to hear the differences. If a low bitrate were used, the enevitable audio artifacts could easily cloud over any differences one was listening for. Ogg Vorbis, even at its mid-point (quality=5, roughly comparable to Muses 'standard' setting), has absolutely no effect on the music's frequency-balance whatsoever. This is a *first* for me. Never before have I encountered a compression system that alters the fundamental structure of the music. Using WinAmp, I've been shifting repeatedly between ogg, muse, and the original wave file, sampling the music at different bitrates, and to my ears, (Lasse), MusePack comes out the loser every time. Sorry. I could go on and on about this...... I have to scratch my head in bewilderment at correspondants here in these pages who ask, 'What's the best settings in Lame for classical music......'. (Or rock, or pop, or jazz, or blues....). It matters NOT what genre of music one likes, - what setting works best for classical should also be the same one that works best for everything else. Bitrate is not directly or indirectly comparable to musical 'type'. The goal here is attempting to get the most honest, unaltered, most-transparent sound that we can achieve as compared to the original. ANY lossy compression system will introduce losses - the trick is to keep those losses to a minimum. Thank you ff123, for the info. Regardless of how much Muses's improvement over the original mp2 format upon which it is apparently based, has advanced, I'll just keep .mp2 use for the audio soundtracks in my white-book vcd conversions. THANK YOU ALL ! (Gearing up now for some extensive ogg vbr compressing at -q5)....... -- KlingonAgent --
Klingon, There must be something wrong with your setup. I mean your findings are false and definitely not caused by the Musepack format. I friend of mine suggested that you might be experiencing this:"maybe he used "EQ controlled by musepack" in the plugin options, and now MPC doesn't use his Winamp EQ settings" If you don't beleive me, then have a look at the thread ff123 posted above.
First off, my personal preference in codecs is mpc, just like Lasse's. But you are correct to trust your own listening preferences. Just a caution that you should check to make sure your comparisons are valid. The best way to make sure is by using the command-line decoder to decode back to wav instead of using the Winamp plugin. Most people who hear problems in mpc (and it's not too many) have been reporting a very subtle increase in brightness on some samples using the --standard profile. If you hear obvious problems with --xtreme, --insane, and --braindead, then that would be very interesting to a lot of people, and also to the developers, who are always on the lookout for problem samples and people who can hear them. I personally would be interested in obtaining sample clips and ABX results from you, using --xtreme if it turns out you are hearing real problems. If you go back to the thread I mentioned above, you can download the castanets.wav test file and see how your results compare with everybody elses. The nice thing about this test is that the encode/decode process has been standardized. That is, there is no chance that some DSP or equalization setting has enabled by mistake, which is more common than you might think. Another common complaint about changes in sound from mpc (people report a dulling of the sound) has stemmed from clipping protection. I haven't really kept up with the changes going on in mpc, but you might want to check to make sure that clipping protection isn't enabled and causing you to incorrectly assess mpc's sound quality. ff123
ff123, I have to point out that LAME --alt-preset standard is my personal preference, since it is way above my personal "transparecy level". I don't have high quality head phones and I am not very interested in listening tests. I don't even want to train my ear to hear the defects of compression. Few years back I did listen and study regular Hi-Fi quite a bit and it cost me very expensive stereo equiptemt to satisfy my need for quality. I am much better in hearing the problems in dynamics, freq. response and such "analog" issues in sound quality, than the problems introduced by digital audio compression. My respect for the Musepack comes from the results of various test results I've been closely watching at the r3mix forum and Hydrogen audio. Also the comments by competing developers (LAME devs, Psytel AAC devs, even OGG devs) support the fact that Musepack is the top performer at the moment.
Hmmmm..... unfortunately there's a finite limit to the amount of time I can afford to play with this stuff. Lasse, your friend suggested, "maybe he used 'EQ controlled by musepack' in the plugin options, and now MPC doesn't use his Winamp EQ settings." Well, I didn't manually configure the plugin, I just dumped the .dll into the c:\progra~1\winamp\plugins folder. (or actually, the installer did it for me). Winamp recognized it right off, and .mpc playback was accomplished. But anyway, I went to the plug-in and discovered that yes, the box under "Equalized By" read "[x] Musepack", not "[ ] Winamp". So, just to waste more time I really don't have to spend, I checked-off the "[x] Winamp" box instead, and was immediately rewarded with a popup: "You have activated 'Equalizing by Winamp'. This will activate WinAmp's generic equalizer. The generic equalizer produces heavy loss of high frequencies!" Gentlemen............. I have no intentions of using a codec's equalizer. I didn't know codecs were *supposed* to be *incorporating* equalizers. Is this a codec, or is this a goddamned equalizer? No other installed codec makes this requiement. All Winamp codecs work as-is, and all use Winamp's standard graphic equalizer to good advantage. I did visit ff123's link, Lasse, yesterday, and read with interest the tests with the castanets. And ff123, despite the wonderful, detailed explanation you have given above, I have not the patience to conduct the indepth testing you suggest, however valid (which it certainly seems). Just an aside though...... Winamp has seen fit to include ogg vorbis as a standard part of it's internal workings, requiring no add-on plugin. Why not musepack? --The Kloth-Eared-Klingon--
I think the Winamp folks are being conservative about including musepack because there might be some patent issues with the codec. I believe Andre (musepack developer) has inquired about licensing the patents from the patent holders, but does not actually have licensing agreements in hand. As far as the mpc winamp plugin and equalization, I haven't played with the latest plugin, so I don't have anything to say about that. However, the first thing I would do (assuming I were you, and were willing to spend a few minutes doing it) would be to do a binary file comparison of the output of Winamp (using the file-write plugin) vs. the output of mppdec.exe to make sure they're byte identical. I presume that some combination of checkboxes will get you to the desired unequalized output. ff123
Klingon, " All Winamp codecs work as-is, and all use Winamp's standard graphic equalizer to good advantage. " The issue here is that WinAmps internal EQ is flawed and will result you loss of quality - no matter what format you use. That is why Musepack plugin includes it's own EQ. Also note that psycho acoustic converion routines expect a natural frequency response (this includes MP3, OGG, Musepack, AAC..) and altering the freq.response with EQ may bring up artefact are are not supposed to hear. But this also depends on your output system, since the final freq.response produced by your speakers is what matters (meaning that flaws of some sound systems may be possible to fix with careful equalizing). Here is a WinAmp EQ that actually works: http://shibatch.sourceforge.net/
Possibly so, ff123, but for now I'm going to stand pat. But you have approached this thing very logically. I've been reading some of your posts, as well as visited Hydrogen Audio and the Musepack Phorum, and elsewhere. I have grown quite fond of ogg vorbis in the interim. There are many things I admire about it. Fundamentally, it sounds superb and uses an advanced psychoacoustical model. (That's part of the blurb describing it). It's a cinch to use, and I expect that was the case even before Winamp incorporated it directly, and required a plug-in. In terms of audio quality, I am totally satisfied; to my ears it surpasses MP3; it creates smaller (30%) filesizes for similar average mp3 bitrates; it's VBR to conserve space; it's totally 100% patent and royalty-free; I've already written batch files to automate it's use; .ogg files can actually be seamlessly joined together (copy /b oggfile1.ogg + oggfile2.ogg bigfile.ogg); an undocumented feature of Winamp--namely--to flawlessly and easily tag any .ogg file it has loaded, is simple and foolproof; it is well-know throughout the Internet & I see a trend in the websites I have visited to introduce the .ogg files over .mp3--even if-- some of those files have obviously been transcoded (.mp3 --> .wav --> .ogg), which drops quality; and SO....... I'll keep an eye on Musepack to see what transpires. Best Regards, -- KlingonAgent --
