Slashdot Mirror
Non-MP3 Codecs?
Vanth Dreadstar asks: "While
MP3 is okay, I have begun researching other codecs that would be
suitable for my home music use. Lossy codecs such as Ogg
Vorbis, AAC,
and MPC all seem to have promise, not to mention the lossless codecs
such as Shorten
(otherwise known as .SHN),
LPAC, and FLAC.
I would like to know what non-MP3 codecs people are using out there,
and why."
I'm using Ogg Vorbis for a number of reason. The reference encoder, while not perfect, is certainly not bad. The vast majority of the time, .ogg's sound noticeably better than MP3's of the same bitrate.
.ogg files with the track names grabbed from FreeDB. To actually encode, one symply drags the .ogg file to another directory, and the IO slave works its magic.
More importantly, Ogg Vorbis is free of any patents or any other restrictions. I could make a commercial hardware player if I wanted to, and not have to pay any royalties to anyone.
Finally, it integrates nicely with Konqueror's audioCD IO slave. You can simply type "audiocd:/ogg/" in Konq's location bar, and it shows you a list of
Lex orandi, lex credendi.
It still doesn't matter to me. If I could listen to WMA on my linux system(s) I would. If I could use WMA on my car mp3-cd player, I would.
I can't though, so it doesn't matter. I'm not a musician by any means, nor can I detect the difference between 160 and 192 mp3 compression. So I'll continue using my inferior, yet cross platform, non-license restricted, used-everwhere, mp3 format.
I use MP3s because they're much like Interet currency.
:)
I convert MP3s to WMAs when I want to squish music onto my PocketPC.
If I bought an OGG car player (if there is/was such a beast), I'd convert my MP3s.
The point: When in Rome, I do as the Romans. It's a simple life, really.
WhatEVA
I can hear the difference between a 128kbps mp3 and the original CD (192kbps CBR or 160kbps VBR are good enough for me), however the difference isn't nearly so great as the difference between playing the music on $30 vs. $100 speakers. You can get decent computer speakers today (if you're not an audiophile and don't need very high volume) for as little as $60, but the prevalence of 128kbps recordings on the internet suggests to me that most of these people are still listening to music on the little white buzzers that came with their computer.
* Picture perfect at 128 kbit/s
I don't know about the rest of you, but to my ears, NOTHING is "picture perfect" at 128kbps. 192 is minimum for any lossy compression.
Counter-rant: So what if "research" has been abandoned on MP3. I don't need that research, 'cuz there are great MP3 encoders already out there. The work has been done.
For archive quality (as opposed to streaming audio), what do .WMA, MP3Pro, Real, and ATRAC offer over 192/256/320k MP3s?
Nothing.
They all support various copy-control schemes, which make for revenue opportunities, which might cause their respective proponents to funnel R&D bucks into them. Some sound better at low bitrates, which is fine for streaming audio, but most folks in the streaming audio are - once again - just trying to make a buck selling pay-per-listen or pay-for-subscription streams.
That's the other reason nobody's researching MP3 -- not only is it "good enough" as it stands, there's no money to be made, even if it could be improved.
Talking about the lack of "cutting-edge research" MP3 as a death knell is like talking about the lack of cutting-edge UNIX text editors as the death knell for vi and emacs.
I don't need Microsoft or Real or Sony to put a million bucks into researching the latest WMA codec, because I know it'll be DRM-crippled and useless to me. The research into other codecs is, for me, wasted. I couldn't care less.
(Likewise, the lack of "research" into cutting-edge text editors doesn't seem to have made vi or emacs go away...)
As for Ogg, as good as Ogg is, I see the odds of it replacing MP3 in terms of the .GIF vs. .PNG debate -- most places that could use .PNGs still use .GIFs, despite GIF's patent issues, because .GIF was "good enough" and widely-distributed before PNG came about.
since then most of [MP3 encoding] happens on cirrus logic processors or TI DSPs.
However, the TI DSPs that handle floating-point arithmetic are much more expensive. Nobody (except Iomega, and even that's not officially released) has made a portable Ogg decoder because the Vorbis reference decoder from xiph.org uses extensive floating-point rather than fixed-point arithmetic.
If you write a Free integer decoder (or fund writing one), they will come.
Will I retire or break 10K?
Easily solved. For, say, the 1.3 point release of Ogg Vorbis, pull a Sun and dub it "Og3". Heck, just call it that right now. Positions it nicely as a competitor to the known format.
If you compare a good mp3 encoding ... to a WMA8 encoding of the same bit-rate and with the volume levels matched, mp3 will win out
I didn't realize WMA8 was compressing levels, but once levels have been compressed, it won't be possible to "match volume levels" and compare with original source or an MP3 as you suggest. (ie, either loud passages won't match or soft passages won't match)
With the quality of the latest RC3 release, Vorbis now sits on the throne in the low to middle bitrates, easily beating out MP3Pro and WMA even in the very low bitrates of 64kbps. The best part about it is that Monty has mentioned that he's still not happy with the quality at 64kbps and will still be improving it further. At middle bitrates of 128kbps, it is at least as good as the best AAC implementation. At the high bitrates, it still hasn't matched MPC, but it is catching up really fast. Whether Vorbis (a transform coder) can ever overtake MPC (a subband coder) quality in the future in the high bitrate arena (usually ruled by subband coders where pre-echo artifacts are nearly non-existant) is very much unknown, and probably depends on Vorbis implementing a really good anti-pre-echo system better than all the current techniques being used.
So therefore, for the best quality now, use Ogg Vorbis at bitrates of 160kbps and below. Above 160kbps, use MPC.
The reason the live music trading community (most notably etree.org [etree.org]) uses the shorten format is because there was not a way to widely distribute exact copies of, say, master DATs. Now, assuming the person transferring the DAT, did a reasonably good job, every person after that who receives the SHN files can create an exact copy of that DAT. This is crucial because of the way shows are distributed. One person gets a copy from his friend, and he passes it on to his friends. If there was a lossy step involved in the middle of the chain, each copy would be worse than the one before. Note tape trading. Copying a...
:)
:)
Hmm... but you're making an assumption here that, for some reason, every person in the chain would re-encode the audio data into said lossy format before sending it to the next person in the tape tree, which would (hopefully) not be true, in general. For example, I try to keep an archive of all the compressed audio files I download, even if I burn them to CD. In fact, often times, I just make a multisession disk with the compressed audio on the data portion. Then again, there's no telling what an uneducated trader might do.
I guess what I'm driving at here is that, if the traders were bright enough not to re-encode all the time, and just pass around the original files, a compressed format could make trading a LOT easier for those with reduced bandwidth. Frankly, I think the community chose Shorten for the same reason some audiophiles prefer vinyl... they think it sounds better (and, IMHO, given the quality of your average taped show, a compressed format probably wouldn't affect quality that much.
With a traditional cassette tree, yes, every member in the chain is adding a lossy generation to the next tape in the chain. That can't be helped with cassettes.
By "original files" what do you mean? Do you mean the original wav files? The original shn files? Or maybe the person transferring the master made some mp3s? Also by "compressed format" are you talking lossy or lossless?
The traders are bright enough to not re-encode all the time, they're passing around original shn files that match an md5sum hash in an established database. That way everyone is guaranteed a good copy; at least they're guaranteed the same quality as the master! The people in this community have a different solution for those with reduced bandwidth: USPS. Mailing around CDs filled with SHNs is still very prevalent.
It is a compromise though. You have to wait a long time to transfer a single show. It's a compromise most are willing to take, though, for the higher quality. Who's going to trade with you if you have a lower quality recording than the next guy? You might not be able to hear the artifacts introduced by mp3, but if the next guy can, he's going to be pissed that you traded him schwag.
The community chose Shorten because they needed a way to guarantee quality. A commercially pressed CD has thousands of "masters". A show taped by the taping community has one, or maybe a few more if he was giving patches. To distribute an exact copy of this music from only one master is quite a feat.
The community also chose Shorten because it DOES sound better. For example, live field recording has a ton of ambience. Lossy compression schemes such as mp3 do not encode that well.