<sss>
size definately is important, to some tech is more important, than again those that say that may have a size that is not appropriate for the task (in their minds) so they need to work on the tech part, so one can conclude that there are obviously at least two ways to get the task uhmm done. but again size-in-time depends on many factors.
</sss>

Well.. just had to put 2-cents in on this one. The young man in this post (which has indicated the American flag) is obviously one of the BIG CORPORATE `Media Propaganda' victims. Americans are SO inundated and confused with nonsense from the TV and NEWS that it's impossible to distinguish REAL from IMAGINARY.

Have you heard of Orwell's 1984?? With `double speak' and `new think'?? This chap would believe that the biggest gas-guzzling SUV is the way to go!! I'm very concerned as to what's happening in the USA .. the corporations could be looting his bank account and he'd go on about BIGGER / MORE .. god help us!!

Actually, he was just joking about budgie's aim for unnecessarily high bitrates...

Oh -- well never mind then .. { grin }

Well, it is very sad, that there are such debates around here... Standart vs. extreme... But I hope, everybody has made the right resume. At least so did I. Shame on me, but I thought that -q5 would be worse then -q8 while listening... But, however, there is a limit to MusePack (and especially to SV7) that is why, I think, maybe for future transcoding to new and better formats it is not unreasonable to encode using q7-8 switch? Or even in that case it is pointless?
And by the way, if I am not going to do something with tracks but just encode and listen to them, is there a sense to use –q4 or –q3? Are they still transparent? If there is no difference in quality between –q5 and –q6 (or even –q8) so why there should be any difference between –q5 and –q4?

Standard is very tuned to be transparent and bitrate-wise.
With q3 or q4 you might hear some artifacts... same comparing standard, extereme, insane.
If you find artifact, then just check if higher setting makes it disappear or less anoying.

I think what "confuses" people is that they can easily hear the sonic improvements as they move from --quality 1 to 2 to 3 to 4 to 5. Since --standard is pegged to 5, some might think that going further - to 6, 7, 8 or whatever - will result in an audible improvement as well. I wonder if this obsession with high bitrate presets have more to do with semantics than anything else...the mere presence of higher bitrate presets makes some people think they really do need them.

This is placebo effect in action... you hear less because you spent more space/effort on this track.
This proves that only blind tests (eg. ABX) can prove anything.

No doubt. Perhaps --standard should really be --quality 9 and the "bitwaste-profiles" --quality 9.2 9.4 and 9.5.

Everyone seems to be debating weather -q 5 is transparent to most people. By now we all know that it is. I use -q 6 because I think it will help me with transcoding later. Where is the three pages of discussion over this?

Are there ABX results from other people between?

WAV -> MPC -q5 -xlevel -> MP3 -aps
WAV -> MP3 -aps

WAV -> MPC -q6 -xlevel -> MP3 -aps
WAV -> MP3 -aps

WAV -> MPC -q5 -xlevel -> MP3 -aps
WAV -> MPC -q6 -xlevel -> MP3 -aps

Some people have made some preliminary testing already. I just want to see more people.

I am always amazed by the amount of people obsessed with transcoding. Transcoding is an enemy of quality and distorts the real value of lossy encoding, which is to give you the most quality for the least amount of space. If you use -q8 so that you can transcode your MPCs, you have given yourself files that are 40-60% larger than they need to be and the transcoded files will always have a poorer quality than if you re-ripped and re-encoded into the new desired format.

You are better off ripping once and encoding each WAV twice by making (1) an "archival" high quality encoding using MPC -q5, and (2) a portable mid quality encoding using LAME MP3 and storing the files separately. This way your MPC files aren't any bigger than they need to be and your MP3s will use the original WAV as the source, not another lossy file.

I don't think this forum should necessarily tolerate the concerns of transcoding to the point where it becomes an obsession and negatively impacts how we use lossy encoders. Quality comes first, not convenience. If you want to transcode, that's your own business but it becomes a problem if we start spreading some "recommendation" that you need to use MPC -q8 so that you can transcode later on. If you want MP3s of your albums, encode them at the same time you encode your MPCs.

Oh, so you think that transcoding is not worthy of discussion here? I would rather transcode then encode every damn song into mp3, ogg, and ac3. But that's okay, I'll just find out this stuff for myself. Then maybe talk about it in a forum devoted only to psychoacoustic audio compression... well, another one then.

LOL

I think torok is right, this is an audio-forum, we should discuss everything related. but U dont have to add your opinion if you think its a stupid topic - simply ignore the word "transcode", eery time you see it - and you'll have a great time

The transcoding subject has come up several time.. but I've declined to comment.. mainly because I do not recommend or do any transcoding. However.. if you wish to transcode .. then go ahead with it .. if you are happy with such a procedure. { statement with a smile }

As stated in another thread .. trandscoding is like repeated saves of a JPG image where each `compress' loses more detail { digital image analogy }. As mentioned by other knowlegable forum members.. I feel it's best to create WAV files from your CDA's .. and do your encoding. Then compress the WAV files with something like FLAC (lossless) as an archive.

If you delete both the WAV and FLAC files.. such as to save disk space.. and another encode is required.. then re-rip the CD. Whatever process you decide on.. have fun and enjoy!!

Hey, I said that! I feel so special.

But really, I don't believe that analogy anymore. I did that research myself, and am firmly convinced now that transcoding lossy sound is the devil. See, with a good quality JPEG, you can resave (transcode) it like 5 times before anyone could even notice a difference. I transcoded a song from MPC Standard to MP3 and it sounded like crap no matter what I did.

And I'm such I geek, I know I'm going to go out and by another hard drive just to have everything lossless now. If only my player supported MPC none of this would even be an issue. Arg!

JPEG is DCT based. DCT is much much more stable concerning reencoding than MDCT
or related transforms. Actually it should be possible to write an JPEG codec which
is absolutely immune against reencoding (without processing).

JPEG2000 (non lossless) has lost this propertity.

I did some tests with a jpeg encoder and the result was constant after 3 reencodings.

The problem with transcoding is that it is not a quality-oriented measure and this is a quality-oriented forum. It's not that we can't talk about transcoding here but I find it troubling that people use or even recommend using lossy codecs inefficiently (i.e. make bigger files than is necessary) so that they can accommodate a future convenience issue.

You say you would rather transcode than encode everything into various formats. But think about it: when you transcode you are still doing this! WAV->MPC and WAV->MP3 requires two steps, but so does WAV->MPC->MP3. And the former is faster because you don't have to decode the immediately lossy file (which is moot for MPC, though, because it can be decoded so quickly).

Well, then let's get rid of the MP3 forum. MP3 looses out in efficiancy at high bitrates to MPC, medium to AAC, and low to OGG. Recomending MP3 is just recomending using lossy codecs inneficiantly to accomodate mere convenience.

I meant that I didn't want to encode to every possible format now, just in case one of them becomes a new standard later. WAV->MPC->AAC is much faster then WAV->MPC->AAC->OGG->MP3. Also, by doing that I'm using about as much space as lossless, and I don't do lossless because of the filesizes.

I did a bunch of ABXing, and at --alt-preset 176 I can't abx an MP3 file, while at 160 I can. It's the same when sourced from an MPC -q5, so I think that is what I shall do. hehe, I'm just discussing it all over the place!

Transcoding is evil, true, but it's not like we're all trying to encode with MPC now and transcode to MP3 128 for archival. Nobody's THAT stupid...i hope.

I'm interested in a transcoding discussion because there are situations where transcoding is the only way to take your audio on the road. I own a cheap 32MB MP3 player (cost 200$ back in the day, arrrgh). I can cram slightly over a half hour onto it, using 112kbps. The quality is fine by me...but only because it's temporary. I delete the MP3s as soon as I'm finished with them. Now I'd personally very well like to know what kind of settings I need to use to achieve "acceptable" quality by transcoding from MPC to MP3, as you can imagine.

I have to agree with mithrandir here. When you look through this forum you will see many people who admit that mpc q5 is perfect, but they may want to transcode in the future so they use insanely (pun!) high presets. Some hope for a new format to come along, which offers the same quality as encoders today at half the bitrate, is platform independant, supported by portable and stationary hifi hardware and makes coffee. Imho it is illogical to expect such a thing. You want to save space in the future (where space will be less of an issue than it is today) so you waste space now? You expect higher quality so you transcode from an inferior source like a lossy mpc file?

You only have to transcode mpc into something else, if you leave the PC environment. There are two scenarios I can imagine.
1: Music for your portable and 2: your hifi integrated mp3 capable dvd player.

When you use your portable, quality is not such a big issue. Lame aps mp3s transcoded from q5 mpcs are still superior to your average 192kbps Blade/Xing mp3s. Less pre-echo, hf flanging etc. The quality of the transcoded mp3s can improve, if you use q6 mpcs as source. Going any higher only has an extremely small effect on the quality of your transcodes.

In your hifi setup, if quality is that important to you, transcoding shouldn't be an option to consider in the first place. Encoding from an allready lossy source degrades quality, no matter how high quality that lossy source may be. If you can live with a little loss, then just use mpc q5 and go from there. It won't be perfect, but mpc q8 won't be either.

Very well, indeed... I see a new industrial standards in the near future: --aps-preset standard for a kind of low-fidelity sound and MPC -standard for high-end users Forget the wavs when MPC -q5 is good enough and sufficient for almost every kind of music

For the "low-fi" portable setting, I'd prefer to have for example, ~80-90kbps Vorbis or something.
I'm waiting for a tiny Vorbis player which would support CompactFlash or SD cards..

After reading some of the transcode topic threads, now I can say I'll begin to transcode, and I think in some situations it's a good idea.



Yes, I understand that coding from a lossy codec to another is going to sound worse that coding from the original CD or wav file, and I have nothing against having the best quality. But let me explain my point with an almost real situation:

I have a near noise-free computer with very good speakers. For having real quality (to my ears) I can use lame --preset standard or MPC -q5. I also have a 128 MB flash portable device in which I use MP3 (--preset 155 sounds fine to me), and my brother has another in which he uses AAC (actualy I have no brother , but this is only an example; maybe in the near future I would buy an AAC portable player and I'll prefer to have the music in AAC 128 instead of LAME 155 ). Besides, I also like listening to my music at work, and in such a noisy environment OGG -q0 does the job, and filesize matters as I work in different locations and I like to move my music from place to place in a 64MB flash UBS disk-key.

Then I have 4 options:

· Ripping all our CDs each time we want to encoded the music with the apropiate codec / bitrate. I don't like this idea because the ripping takes time and because that mean spining our original CDs at 16X tons of times.

· Ripping all our CDs once, the encode everything in MPC -q5 (home PC) OR LAME aps / ape AND LAME --preset 155 AND AAC AND OGG and delete the original WAV files. This is the most stupid posiblity as I don't have so much HD space, and the total size of each song encoded 4 times can be vey near the size achieved with a lossless encoder.

· Ripping all our CDs once and encode them with a lossles codec (FLAC, APE ...). Then I can code from a lossles source each time I need without losing quality. This is by far the best idea, but 250 CD's x 400MB/CD = 100GB !! I don't have enough HD space and another HD doesn't fit in my case

· Ripping all our CDs once and encode them with MPC -q6. Then I'll be able to code from a lossy (but transparent) source each time LOSING very little quality (from my point of view). As 250 CD's x 93 MB/CD = 23 GB !! I don't need another HD !!

Can you see my point. It's all only about changing quality for space. I thought the psycoacustics compression methods are about that. Each person and each situation is different, and for some of them transcoding LOSING quality can be no such a bad idea.

Just a thought...

Glauco,

I understand what you're saying, and agree entirely. I rarely have to transcode, and if I do, it'll be a few tracks at most. In this case, the advantage of easily finding these tracks using an mpc from HDD (rather than finding the CDs and ripping them again) far outweighs the disadvantage of using a few more bits for q6 over q5.


For me, it's usually (but not always) a question of giving music to other people. (Out of copyright music, you understand, and I'm talking hypothetically). They usually want an mp3. I'll explain about mpc, but they're not interested - they want an mp3. A 128kbps mp3. That's exactly what they want.

I could

a) spend lots of time explaining why mpc is better, or why lame --aps is the best mp3 etc etc
b) find the original LP (seriously - if it's on CD, they can go and buy it! ;) ) and make another copy, and encode it to 128kbps mp3
c) take whatever digital copy I have handy (mpc q6, for arguments sake), and encode it to 128kbps mp3.


Since I've already offered them a higher quality version, but they said no; why waste my time? That 128kbps mp3 is going to sound fine to them. And if it doesn't - they can come back and learn about mpc!

In this hypthetical situation, I may have a lossless copy somewhere. In that case, I'll encode from that. But if not, then I won't re-rip or re-record. Unless they are bothered about quality, in which case they can have whatever version they want, taken from the original. But most people aren't bothered.


fwiw, I totally agree about the "encoding at q9 so I can transcode to a better format later" - it's silly! But, hey, it's your life and it might work, and you're not hurting anyone. But which is more likely:
a) this really wonderful new format comes along - it's open source, it has great hardware support, and it transcodes perfectly from mpc -q8
or
b) you wish you could fit 30% more tunes on your HDD or each CD-R - so you transcode mpc -q9 down to mpc -q5 - which results in sub mpc -q5 quality, but slightly higher bitrates.

I bet there's a couple of folks out there heading for (b) very soon! ;)


So, what's the difference? Well, if you're building in a safety margin because it'll make your life easier now, with encoders that exist currently (so you can check what actually happen), then there's some sense to it. If it's some "castles in the sky" plan to future proof your lossy collection, whatever comes along, then I think it make less sense. Just store the originals safely, be happy with your mpcs, and re-encode from the originals if something much better comes along.

Cheers,
David.

To start things off.. Transcoding is a very bad.. And will only make files sound worse. However there are cases where isn’t too bad, allow me to explain:

My collection has a lot of MPC files in it, I use them for backup and casual storage for listening in my room. I also have a Kenwood MP3 player in my car. Now it’s nice to be able to burn a bunch of songs to a CD(data format) and listen in my car. The only way I could do this without transcoding is to just use MP3 exclusively. This would be ok I guess, but then I would lose out on all the quality gains of MPC, not to mention it saves me space on my HD.

So what do I do? I transcode.. MPC q5 -> MP3 alt standard.. Everything still sounds good to me, I hear no artifacts that wouldn’t be there anyway if I made the MP3 from scratch ( I only notice this on certain songs I’ve ABX many times).. Also note that I take crappy care of these CDs because I get bored of them quickly and I throw them out after a month or two. So in the end, these transcoded files never make it back to my database of music, and I have nice high quality MPC files to hold on too.

I've searched and, for all the times this is discussed, and all the times that people say "it's bad - don't do it", I've not been able to find much evidence of people actually listening. (please link if I'm wrong - "transcoding mpc" brings up little).

I just tried the few test clips I have on my PC,
wav > lame --alt preset 128
and
wav > mpc --quality 5 > wav > lame --alt-preset 128

About half sounded horrid at 128, but the verison which went via mpc sounded no worse.
There were two samples (Fugue and jump_long) where the version which went via mpc sounded noisier.
The others sounded "OK" at 128, and sounded no worse on the transcoded versions.

From that, it would seem that, MAYBE, some tracks will suffer, others may not.

I tried the same thing with --alt-preset standard -Z, and on first listen I could detect no ill effects from transcoding via mpc compared to encoding directly. When I listened again, I found Fugue had more noise behind the notes in the transcoded version. I'm fighting with an SB-live today, so can't ABX properly (the program won't run correctly), so can't say more than that.


There must be a point where you can say that 99% of tracks (for example) will transcode without harm to mp3, but I wonder what --quality that is?


Accepting that it's a bad thing to do, can we discuss/investigate it?

Cheers,
David.

Budgie, I get the impression that you believe the quality of an encoding improves steadily the higher you take the settings. What we are all saying is that this IS NOT THE CASE. Past -q5 or perhaps -q6 you will NOT HEAR ANY IMPROVEMENT in audio quality no matter how high the bitrates go. PLEASE, for the sake of rational thought everwhere, download winABX, and run a normal and braindead version of your favorite song through it. It'll take you 10 minutes at the most, and I don't thing you will be able to tell the difference at all. If you can't, why not switch? If you can, keep it up! Conserve those bits, and send them to the poor people of Uganda or something.

Transcoding from MPC to MP3 (especially at bitrates as low as 128kbps) is not EVIL at all, since quality loss for trancoding from MPC (subband-codec) to a transform codec (Vorbis, MP3) is not as big as transform to transform.

dev0

well said indeed..

This seems to be common knowledge, but if I remember Frank Klemm's words correctly on this part, then this isn't neccessarily true.

Some time ago (mpc 1.0x era perhaps) I did some transcoding tests from mpc q5 and q6 to mp3 aps. I posted the results in a response to a thread started by liekloo, but sadly they were lost during the board software migration. IIRC I could hear a difference in about 1/3 of the q5->aps files but none with q6->aps.

However I happily transcode from q5 and up files to apfs for portable use and have almost never stumbled upon an obvious artifact (and if I thought I did, I never investigated... it might have been bad mastering from the start). When you go as low as 128kbps then I wouldn't worry anyway.