i am just curious. In in almost every post i come arcoss that has the subject of "which one is higher quality .mp3 or .mpc?"
The vote seems to always go to the musepack format.

So, i am just looking for some information on why/how is .mpc better the .mp3. I have hear the term Subband encoder, but i dont have a clue to what it means, and what the benifits of it are.
If any one could point me in the right direction on where to find some information on the .mpc format and what makes it so good.

Im looking for a lossy approch to achive "archive quality." (that funny, lossy and archive in the same sentence. but you know what i mean.) Currently in been using lame -ape to encode my music. And the quality seems to be great. Im just thinking about trying different formats, but i DONT want to lose quality. So if i was to switch to the .mpc format im guessing that i would be using the high-end swiches -insane, -braindead.......


thanks
-ty1er

Hey,

I don't really have the technical background that some guys here have, but I think I can give you some reasons that explain why mpc is better than mp3.

- MPC is a younger format and developed with quality in mind (actually because the author didn't find mp3 having enough quality I thought)
- it is a subbandencoder wich I thought was always better in the higher (>160) bitrates than transformencoders (don't shoot me if this isn't true, it's just how I understood it )
- it is much more finetuned and mature than many other formats out there that are still evolving

As for what you should use, I also used to encode everything with lame- ape but changed to mpc because with that I could achieve more quality at the same bitrate wich makes more sense to me.

I even think that with mpc the standard profile will be the same if not better quality then lame -ape.

BTW I encode with --insane --minSMR 0 --nmt 12 --tmn 28
Why such a commandline you ask?? I just wanted a commandline that gave me around lame -ape bitrates, so I ended up with this one and I never regretted it because the quality is GREAT.
I know that I probably could get away with extreme or standard but this commandline gives me a good feeling of having some spare bits for the hard to encode material. (It's all in my head I'll tell ya )

I don't have any practical experience with MPC, but for what I've read here, MPC --standard has quite better quality that any mp3 preset at any bitrate will ever have.

If you want "the best" quality in lossy format, MPC insane is your only choice. If it do come to a date you want to transcode it or anything.

Insane was never optimized for best quality.
It was only optimized for people using cooledit
to evaluate quality of Codecs.

--standard normally enough
--xtreme much better hearing than me,
transcoding
--braindead much better hearing than me
2...3 transcodings should be possible without serious degrading audio

--insane invest a lot of bitrate in the HF range,
--braindead tries to invest it over the full
spectrum

Take --braindead. This is more secure.

Okay Frank, I will give braindead a try but I thought (I read it somewhere) that the braindead profile was going to disappear again in later mppenc versions because it was a bit silly to use and was only to show that there are profiles even better than insane.

Could you give me an idea of what is in the braindead switches??
I mean nmt and tmn ??

--braindead almost represents this:

--insane --minSMR 0 --nmt 12 --tmn 28 --ltq_gain -12 --ltq_max 65

"Almost" because this line produces files with a slightly higher bitrate... there's some more tuning for mid/side stereo (2.5/12 dB) which i couldn't figure out now.

Hmm, I´m always reading that --insane isnt´t optimized for the best quality because it uses too many bits for encoding the full frequency range.
But we never get a clear statement on how --insane compares to the other profiles.
I´ve been encoding many albums with --insane and would really like to know how the quality compares to the --xtreme / --standard profiles.
Is insane worse than xtreme? Or even worse than standard?
I´ve asked buschel about this topic and he told that --insane still is of better quality than xtreme.
Now I just wanted to ask you, about your opinion about insane, cause you don´t seem to be convinced about it.

Don´t understand me wrong. I love MPC and I have no problem with the quality. It´s only that this insane-issue has been floating around the forums that long and the opinion of one of the developpers would be quite interesting.

Sorry that i answer this, but i already know what the answer will be. --insane wastes more bits in the mostly inaudible region of 18-22 KHz. However, --insane can't be worse than --xtreme for the region of 0.02-18 KHz, it uses higher NMT and TMN values and less aggressive ANS.

At least the frequencies 20-22 KHz are wasted bits with plain --insane. Only good for transcoding maybe... That's why --braindead utilizes "--minSMR 0", to enable the lowpassing.

the insane switch will output the best quality of any profile...except the new braindead profile which is quite silly...

think of the insane profile as the exteme profile plus extra information to encode the full bandwidth. [i know the differences, please don't post them]

it goes "thumb < radio < standard < extreme < insane < braindead" for the quality verses the actual profiles...

but if you really want to use the braindead profile, think again and use "--insane --minSMR 0 --nmt 12 --tmn 32" without the whole bandwidth, or "--insane --nmt 12 --tmn 32" with the full bandwidth.

you don't need more sensitive ltq than insane for any reason.


later
mike

btw...

no profile will represent the best quality [not like api for lame] with musepack...it would look something crazy like

--insane --nmt 99 --tmn 99 etc etc.

the musepack format can exceed 1mbps, but as a lossy encoder, this wouldn't be ideal in any situation...right citay? lol

just kidding...


all documented evidence points to --nmt 12 and ltq fil being sufficient to handle artifacts, but that was a while ago... and klemm being as bad-ass as he is has probably brought the musepack format to even greater abilities...

later
mike

I heard that "--braindead" is probably the best profile.
But when I encode to "braindead" the file size is less big than "--insane". :confused:

I would also say that --braindead is better than --insane. --insane, as i said, wastes bits in inaudible HF regions. When you would substract those bits, --braindead is more accurate for the audible region. Maybe --insane needs a fixed lowpass at 20 KHz... even the "Cool Edit"-fanatics should agree that 20-22 KHz is completely inaudible to the human ear, for all we know.

@xmixahlx :
Does the --tmn 32 make a big difference from my --tmn 28 ??

I always thought that the --nmt value made more difference than than the --tmn one?? (not sure about this though)

It's just that I encoded everything till now with my tweaked insane line and this has become like a standard to me.
So would it be woth for me to change to the --tmn value you suggested??

I know I should do some listening tests but with mpc it's just impossible to find anything that doesn't sound transparent. Maybe I'd use the higher --tmn value cause I'm going to upgrade my soundset in the near future, but I doubt I will then be able to hear the diiference.

--braindead do really often encoded up to the maximum frequencies you found
on the source medium (often 19.5...20.5 kHz, seldom 22 kHz). The ATH is limited to
a really low maximum level (24 dB less than --standard).

--insane also stores the quantization noise of the CD between 20 and 22 kHz
(often there's no signal there) with full resolution.

--insane makes also those happy who are listening through Cooledit.
--braindead is really insane from the point of encoding margin, but it uses
the bits where a Diamond ear may have a benefits from it.

Ok Frank,

This is as I understand it :

to have perfect audible and inaudible audio quality

-1- standard
-2- xtreme
-3- braindead
-4- insane

to have perfect audible audio quality

-1- standard
-2- xtreme
-3- insane
-4- braindead

Let's say that with perfect I'm referring to transparentness

-=Ducky=- also got his Cool Edit glasses on today, hehe...

I'm gonna make it real simple: --braindead provides better quality than --insane. In reality, you probably won't hear any difference. But "inaudible audio quality".... come on, are you encoding for your dog or for yourself? :listening ARF!

Hehe nope I'm only interested in the audible part believe me but just to make things clearer I made those little lists.

This also clarifys where all those bits with the --insane profile go into.

I already knew that, that's why I'm using --minSMR 0 but I never knew that --braindead was also using --minSMR 0, I thought it also encoded everything but with more quality than --insane somehow.

This was a very good thread to read, I have bookmark this thread. My only question is, is there a site or something that i could read to get all this info. I'm sure every question I have has been answered at one time or anoughter on this forum. The problem is finding exactly what I'm looking for. Everybody has their own opion about what is best, irregardless of if i can hear it or not.
Thanks Klemm for clearing alot of the misinfo up. Stupid Question......I'm not exactly sure what transcodeing means....Told you it was stupid.
If I'm using speeks frontend and decode I now have a .wav file and a .mpc. If I burn with that wav file then delete it is my .mpc still intact as it was or is it degraded?

Transcoding is the act of converting one format to another. It usually means loss of quality, but with some of the command lines presented in this and other threads, I doubt you'll ever hear this loss.

An example of transcoding would be if you had a collection of MPC files you wanted to use them on your portable MP3 player. The MP3 player surely doesn't support MPC, so you'd have to transcode them first.

The MPC won't change, don't worry.

Case is correct, you would only lose quality if you Reencoded that .wav back into a .mpc/mp3/ogg/etc file.

-ty1er

Once again This forum has come thru.
Thanks Case and Frank and all else involved. If there is any thing I could do to contribute to Mpc let me know. I'm not a programer but have plenty of time to listen, test.....ect.