Are all mp3 encoder/decoders equal?

[davelivinstone]

I'm bumfuzzled. Yesterday, I was mixing down a live jam session that was done about 5 years ago on a Tascam Midistudio 688. The final MP3 sounded pretty good on my computer. I put it in my DVD player and it sounded like crap. I took it to my friend's computer and it sounded good. It sounds good on my computer at work. I put it in my portable mp3 player and it sounds like crap. What's the deal? Is there a huge difference between software and hardware decoders? Or does the problem lie elsewhere? The other MP3s that were from the net sounded ok on my DVD player. Any ideas?

The live recording is located at

www.nowhereradio.com/davetestarea/singles

the song is TASCAM.MP3.

Thanks for any help offered.

[TheRealWaldo]

Is there a huge difference between software and hardware decoders?

There is a huge difference between encoders/decoders period. What it's encoded with will affect how it sounds going through different decoders, and how close to the standard it is.

What did you use to encode this tune?

W.

[davelivinstone]

Hi Waldo,
I used the encoder that comes with CoolEdit. I think it is "MP3me". I have a different encoder that I downloaded, SCMPX. Maybe I'll try it. What encoder do you prefer?

Thanks for the info.
Dave

[matty_boy]

I use LAME and it works really good for me. Plus, I think that you can still get it for free.

[TheRealWaldo]

I'll agree with my provincial neighbour, LAME is the best encoder around for overall performance, and yes, it is still free.

W.

[matty_boy]

Waldo, Manitoba or Alberta?

[NickHall]

I'm in danger of getting shot down again, but we discussed this in a LOT of detail here:

http://homerecording.com/bbs/showthr...threadid=55283

Basically LAME is probably the best, and can get fantastic CD 'Sounding' quality mp3's (Yes I did add the word 'sounding' in there...

Peace

Nick

[TheRealWaldo]

Hehe, that's okay Nick, I think you got my point

matty_boy> AB

W.

[ZEKE SAYER]

hey waldo,

i think something may be wrong with your site man. 9 out of 10 time i will go to your website to play a song and the real player won't let me. i have to play the song lo-fi before i can hear any thing. but all the other websites(mp3.com/ect) work fine. what do you suppost that this would be?

[Pinky]

Lame using member here...

Never thought I'd admit to being lame, but hey

Oh, and I use "razorlame" for the interface, has TONS of VBR features

[Pinky]

http://www.dors.de/razorlame/download.php'

grab razor lame there

Also, for anyone using this software who hasn't donated something, shame on you!! I am now the 7th official person to do so!

I suggest you part with a few bucks (I cleared my paypal to him) to support the product and their hard work (and GOOD work mind you).

[participant]

I use the Fraunhofer IIS encoder "l3enc" exclusively... Fraunhofer Institut (Germany) invented mp3... I think they know what they're doing Middle-side encoding only... 128kbps... IMHO 128kbps/middle-side out of the l3enc sounds as good as 160kbps out of any other encoder I've heard.

Just don't use any encoder based on the Xing MPEG codec

If anyone's interested, I wrote a "front-end" for batch encoding with l3enc (not a lot of options installed, but it can automatically add the v1 tag -- no v2 support yet)... did this a while ago. PM me and I'll send ya a copy of it.


Chad

[TheRealWaldo]

Zeke, send me a screen shot of the error messages from realplayer, and copy the URLs (right click, copy shortcut) that don't work into an e-mail to me.

RealPlayer, BTW, sucks.

W.

[TheRealWaldo]

participant:

You know that they can legally come after you for royalties, on every mp3 ever downloaded of your, plus that codec is not free (you are probably using a crack), right?

W.

[participant]

could be. I downloaded a shareware program front-end for l3enc about 3-4 years ago... it did a poor job, so I wrote one that did a better job.

Now that I think about it, they probably didn't have rights to l3enc either. I just assumed download.com wouldn't allow those types of illegal shenanigans. But, that's where the l3enc came from.

Recently went back to Fraunhofer's site to look for info on L3, but couldn't find the pages I had seen before, about how 128kbps m/s should be good enough for all encoding purposes. In fact, I couldn't find anything on L3ENC on their site anymore.

If it were possible to pay for it now, I'd do it.

At any rate, I own a legit copy of CoolEdit 2000 (only use for editing... not tracking) and it includes a lite version of an IIS mp3 encoder. It's not nearly as fast (takes about 20 minutes per 5 minute file), so that's why I'm still using the old. IMO it's well worth the $$, if only I could find out where to pay for it on their site.

[participant]

just to show I'm not being disingenuous:

Here's Fraunhofer-IIS' product page... apparently, all support for L3ENC is discontinued... MP3ENC or FastEnc are the only encoding products seen at this stage, and I haven't tried either.

[TheRealWaldo]

L3 Enc is actually the designation for their codec (Mpeg Layer 3), not the encoder. (L3Enc also used to be an encoder that they developed, but they renamed it due to conflicts, long ago) MP3Enc has been around since near the beginning, and is the actual command line encoder that was created by Opticom to interface with the Fraunhofer codec.

Their basic encoder is only capable of producing CBR format mp3's. The last modification to the software, let alone their site, was early 2000.

Thomson Multimedia owns the MP3 format created by Fraunhofer, and if you use their encoder, each song your distribute in their format is subject to a 2 - 3% royalty, payable to Thomson, of any income you make.

Some additional information resides at:

http://mp3licensing.com/royalty/index.html

In otherwords, spread your money around.

W.

[participant]

Originally posted by TheRealWaldo
L3 Enc is actually the designation for their codec (Mpeg Layer 3), not the encoder. (L3Enc also used to be an encoder that they developed, but they renamed it due to conflicts, long ago

Hmmm. I must have the old encoder, then (4 yrs old, or something) because it's definitely "L3ENC" (at least that's what the command line window pops up as)...

Thomson Multimedia owns the MP3 format created by Fraunhofer, and if you use their encoder, each song your distribute in their format is subject to a 2 - 3% royalty, payable to Thomson, of any income you make.

So, 2 - 3% of 0.00 is still 0.00 Wasn't really planning on mp3.com, or any "income" sites at all; possibly garageband.com and definitely your site, Waldo.

Thanks for the info I'll make sure to pay the people if ever I decide to put up an mp3.com page, or something.

[regebro]

The way you decode a stream is fixed, hence the difference between different decoders are minimal, because all decoders are supposed to decode the same data into the same wave.

Some of course have bugs in the decoding of certain data, but this seldom produces any audible artifacts.

Encoders however, is something completely different. There are NO RULES WHATSOEVER to how to convert a wave into a mp3 stream. You are simply allowed to do this to the best of your abilities, so that the resulting decoding will sound as good as possible.

This means that different encoders will have completely different schemes on how to produce the mp3 stream, which will result in different sounding mp3's.

The best encoder is Fraunhofer, but beware: It has lots of different settings, so each program that uses it will produce slightly different mp3s. I don't know which program to use to get the best results.

[CrankyOne]

From what I've heard, Lame is the best when you use bitrates larger than 160 or so. If you want to have decent quality but keep bitrate low, then you could use L3Enc. Of course, decoding the mp3 as VBR always helps to get the file-size down.

[muzeman]

I have Scmpx and Razor Lame.
Both work fine.

I think if your posting on Nowhere,you don't want to encode higher than 160 anyway.
Waldo would be the one to ask on that.


Pete

[TheRealWaldo]

'The way you decode is fixed'

Incorrect, as there is different encoding algorythms, there is different decoding algorythms. There is accepted standards, but you can get very different results from one codec to another.

W.

[regebro]

Nope. There is one and only one algorithm. Howvere, ther are many differen implementations, and several of them have severe bugs. But it's not a question of choosing different ways of doing it, as with encoding, but of bugs. A specific mp3 stream should always result in the same raw bit stream (with the exception of output errors).

See: http://privatewww.essex.ac.uk/~djmro...ers/intro.html
for more info and extensive test data on tons of decoders.

[TheRealWaldo]

If every algorythm was the same, there wouldn't be errors in some over others! The CBR/VBR algorythm is completely different, etc.

There is many subsets of encode, which require different algorythms to decode, and some decoders fall back on older algorythms etc.

W.

[regebro]

Originally posted by TheRealWaldo
If every algorythm was the same, there wouldn't be errors in some over others!

Yes there would, because as I said, there are errors in the implementation of the algorithm. One algorithm can be *implemented* in different ways. There differences in decoding an mp3 stream is because of errors in the decoding implementation. Not because of differences in the algorithm.

The CBR/VBR algorythm is completely different, etc.

Yes, just as the algorithm for decoding real audio is completely different from the algorithm for decoding mp3, because the data is stored differently.

There is many subsets of encode, which require different algorythms to decode, and some decoders fall back on older algorythms etc.

Did you look at the link I posted?