Quality of MPC to MP3 ?

Topic: Quality of MPC to MP3 ? (Read 5595 times) previous topic - next topic

0 Members and 1 Guest are viewing this topic.

Quality of MPC to MP3 ?

2001-12-10 14:29:11

This is the situation I'm in:

I want to encode my CD collection to my hard disk, as transparent as possible (I also want to write collections or copies to CD-R, using the encoded files as source).

I also need MP3 support (for my AVP Soul MP3/CD player).

My original plan was to use Monkey's Audio for the archive on hard disk, and create MP3s from these files when I need them. Because Monkey's Audio is lossless, this would give me the same quality as if I ripped the tracks from CD and encoded the .WAV files.

However, I decided not to do this because I won't have enough disk space for the entire collection.

Now I'm thinking of using MPC (insane) for encoding the files, and also as a source for creating MP3s.

With Media Jukebox, it's fairly easy: just setup lame as an external encoder, set the command-line switches (I'll probably use --alt-preset standard), and select the "convert" command.

However, since MPC is not lossless, I'm a bit concerned about the quality of the resulting MP3 files. They don't need to be absolutely transparent (my portable player isn't really "high end" audio equipment anyway), but still have fairly good quality.

I'm going to do some listening tests to check the quality, but meanwhile I'm wondering if some of you have already tried this and can comment on the quality. How much impact does using MPC instead of .WAV as a source have on the quality of the resulting MP3 ?

Thank you in advance.

Quality of MPC to MP3 ?

Reply #1 – 2001-12-11 04:28:53

Personally I NEVER transcode. If I were you, I'd get another hard drive and stay with the Monkey's idea.

Quality of MPC to MP3 ?

Reply #2 – 2001-12-11 18:56:46

I would have to agree with layer3maniac to an extent on this however lots not be hasty. If youve got the resources to go the lossless route more power to you, but i can say with relative ease that using mpc as your archive is still not a bad idea. It may be a lossy codec, however, i think i can speak on behalf of the community that it is probably the most transparent lossy codec that most of us have encountered. If it makes you feel any better, even Dibrom has considered going the route you are thinking... if you ask me, if its good enough for him, then its good enough for me and thats coming from zealous audiophiles The best way to decide is go with your ears... encode a few different kinds of music from your collection and compare... if there is a noticeable difference go with what works better however unless you have ears like Dibrom, its unlikely you will be able to differentiate mpc versus monkeys. But then again... if youve got the resources a little headroom can never hurt.

Quality of MPC to MP3 ?

Reply #3 – 2001-12-11 19:55:41

Quote

i can say with relative ease that using mpc as your archive is still not a bad idea. It may be a lossy codec, however, i think i can speak on behalf of the community that it is probably the most transparent lossy codec that most of us have encountered. If it makes you feel any better, even Dibrom has considered going the route you are thinking... if you ask me, if its good enough for him, then its good enough for me and thats coming from zealous audiophiles

I thought we were talking about transcoding as opposed to archiving here. Personally, I don't even use lossless for archiving. I just backup wavs to CD's. A nearly transparent lossy codec is still a lossy codec. Are you saying that Dibrom transcodes from mpc to mp3? That's news to me. Has anyone done any testing of this? If I get the chance, I'll do some testing myself today.

Quality of MPC to MP3 ?

Reply #4 – 2001-12-11 20:10:20

I'd have to say if you can go the lossless route, then by all means, go for it. However, the question was whether or not there would be a noticeable quality difference going from high quality mpc to mp3 vs lossless to mp3. I believe the answer will likely be no, to very little. I have transcoded some in the past and things usually turn out fine, especially if you use the -insane mode. Keeping in mind that you already made the concession that you are not looking for absolute transparency (because you know that is unrealistic) for your portable, I think that's all the more reason to go with whatever is more convenient.

About my ideas on transcoding.. in the past I have used modified MPC settings that produce bitrates near 300-400kbps and use these for "archival" purposes. I don't really have any qualms about transcoding to MP3 from that source because it is much more likely that any artifacts added to MP3 in the process are not going to be significantly worse than the many artifacts that already exist as a given in a format like that. And that's assuming there would even be any real increase in audible artifacts.

Like the original poster, I use MP3 mainly for portables, the quality is not high enough for me to consider using for archival or near archival purposes in the first place.

That's not to say that MP3 "sucks" either. Just that even with my alt modes, at least half of the music I listen to which is transient and impulse heavy, is still far from transparent even though it may sound better than with just about any other MP3 setting/encoder out there.

In addition to all of this, if you are going to be transcoding from a lossy codec, MPC is certainly the codec to transcode from. Due to the design of the codec (partially in that it is subband instead of transform) you aren't going to have problems with dropouts or temporal smearing made worse in the transcoded files due to their presence in the original encoding. With mpc -insane you have a minimum signal to masking ratio of 3 db which means that all frequencies are at least encoded to such a degree (no possibilities for dropouts period, something which according to Ivan, not even AAC can claim), and the temporal resolution is so excellent that pre/post-echo are non-existent.

Quality of MPC to MP3 ?

Reply #5 – 2001-12-11 20:26:56

Thank you all for your feedback, I appreciate it.

BTW sphoid, I like the photo. The Holy Grail is one of my all-time favorite movies.

"THERE ARE SOME WHO CALL ME........... Tim..."
"Death awaits you ALL with nasty, pointy teeth !!"

Heh. Sorry, I got a bit carried away

Quality of MPC to MP3 ?

Reply #6 – 2001-12-12 05:30:51

Quality of MPC to MP3 ?

Reply #7 – 2001-12-12 06:48:34

Isn't that the movie in which they are riding the stick horses?

Quality of MPC to MP3 ?

Reply #8 – 2001-12-12 09:41:42

Dibrom,

Quote

.. in the past I have used modified MPC settings that produce bitrates near 300-400kbps and use these for "archival" purposes.

Could you share this modified MPC setting with us?

Quality of MPC to MP3 ?

Reply #9 – 2001-12-12 12:36:06

Quote

Isn't that the movie in which they are riding the stick horses?

Not even stick horses. The knights are just pretending to be riding a horse, while their squires are knocking together coconut halves to imitate the sound of galloping.

The Pythons got this very funny idea because they had no money for horses.

Quality of MPC to MP3 ?

Reply #10 – 2001-12-18 12:13:52

Quote

Originally posted by Dibrom
Due to the design of the codec (partially in that it is subband instead of transform) you aren't going to have problems with dropouts or temporal smearing made worse in the transcoded files due to their presence in the original encoding.

I'm not sure what you mean by this. I could implement the analysis part of most codecs either as a transform or filter bank. I'd say the difference is more the large overlap used in the MPC analysis/synthesis part. Could you provide more info?

Quality of MPC to MP3 ?

Reply #11 – 2001-12-18 14:51:24

I'd like to apologise in advance for not reading the thread, in case somebody already boldly stated this, but perhaps you should use LPAC instead of Monkey's Audio. It's my impression that it gets lower bitrates.

Quality of MPC to MP3 ?

Reply #12 – 2001-12-18 15:28:49

Quote

Originally posted by petracci
I'm not sure what you mean by this. I could implement the analysis part of most codecs either as a transform or filter bank. I'd say the difference is more the large overlap used in the MPC analysis/synthesis part. Could you provide more info?

I'm not talking about the analysis so much as the coding. Seperating the frequencies into subbands and then quantizing them directly as opposed to transforming them first, offers a much higher time domain resolution (I'm sure you've read the link on the front page by now since you posted a comment in response). This higher time resolution allows for less temporal smearing in the form of more classic echo, and less smearing of quantization noise from frame to frame which can create dropouts like you see in some transform codecs.

Also, at least in the case of AAC (MP3 is not even in the equation), according to Ivan, the filterbank design itself of this transform coder makes it difficult to impossible to *entirely* eliminate dropouts. MPC can (and does) do this easily, even up to 22khz if user wants. I believe Vorbis would be similar to AAC here, but I'm also not entirely sure on that. I've discussed this issue with Monty some but I'm sure whether or not it's possible at all, or whether it's possible without huge bitrates like 300kbps+. MPC seems to be able to encode all the way up to 22khz with no dropouts with about a 30-40kbps overhead (comparing xtreme mode - ~190kbps to insane mode - ~230kbps) for a minimum signal to mask ratio across *all* frequencies of 3db (you can adjust this).

I think for these reasons, MPC lends itself much more to transcoding than a typical transform coder.

Quality of MPC to MP3 ?

Reply #13 – 2001-12-18 15:31:17

Quote

Originally posted by NeoRenegade
I'd like to apologise in advance for not reading the thread, in case somebody already boldly stated this, but perhaps you should use LPAC instead of Monkey's Audio. It's my impression that it gets lower bitrates.

Hrmm.. do you have some evidence to back this? I'm under the impression that Monkey's Audio is about the most effective lossless compression available, though I choose not to use it more for compatibility reasons.

I don't think I've seen LPAC beat APE so far, but I could be wrong since I don't use either very extensively. Most of the data I have seen though has shown that APE is superior.

Notice