IDM Mailing Archive

One important factor many seem to miss is that MP3 encoders are not equal. The best-sounding are probably LAME and Fraunhofer, and the worst probably Blade and Xing. These all sound different from each other at the same bitrate. They are also not equal to themselves. LAME is an ongoing project -- they are up to v3.93.1. There are several Fraunhofer encoders that do not behave exactly the same either. Plus, LAME and Fraunhofer both have settings to trade off between sound quality and encoding time. Fraunhofer is relatively slow, and LAME relatively fast. In terms of overall character, *very* generally, Fraunhofer can sound artificially crisp, while LAME can soften very crisp or sharp sounds. A web search will locate a few tests people have done, and some test cases that encoders can have trouble with. If you'll forgive a bit of self-promotion: http://www.burningrome.com/music/sounds Most of the MP3s are encoded at 128kbps using one of the Fraunhofer encoders. The choice of 128kbps is for small file size -- quicker download by modem, and less disk space on the shared server where storage is limited. The choice of the Fraunhofer encoder is for acceptable sound at a relatively low bitrate. The tradeoff was longer encoding time. A few days ago, I posted a 76-minute live performance I did for WNUR 89.3 FM Evanston/Chicago in September, 2001. Those MP3s are encoded at 160kbps using LAME 3.93.1 at highest encoding quality (-q0). The files are a reasonable size, encoding was quite fast, and the sound quality is functionally equivalent to the original recording -- especially as compared to the FM radio broadcast. If the idea of "ambient digital brutality" sounds vaguely interesting, please grab the MP3s and enjoy. --Mark

quoted 1 line

>>>

To: BAlbers@premiereradio.com From: "alan flood" <guile133@hotmail.com> Cc: idm@hyperreal.org Subject: [idm] ATP (ae) track and crappy MP3 encoding Message-ID: <Law10-F29lrg2nRySLp000078ff@hotmail.com> Speaking of encoding mp3's .......... who the hell encodes at bitrates lower than 192?????? Soulseek and the like are filled with tracks and albums that are encoded at 128 and 160. Pretty moronic if you ask me......even more moronic are the people who download these low quality tracks and actually listen to them. It should be a law that you mujst encode at 192 or higher. I guess some people like listening to there tracks with the bandwith noise <<< --------------------------------------------------------------------- To unsubscribe, e-mail: idm-unsubscribe@hyperreal.org For additional commands, e-mail: idm-help@hyperreal.org

Mark, Just a quick thing to point out... generally Fraunhofer is MUCH quicker than lame.... (my lame encodes sometimes get down as low as 2x speed on my 2100+ XP machine.... fhg codecs are close to MPC with encoding speed at around 18-20x or more with my machine) its not recommended to use the -q settings alone, rather the --alt-preset settings are optimised fully including changes not exposed via traditional command line switches... --alt-preset encoding is VERY SLOW but very good quality... with lame you HAVE to use VBR... not ABR or CBR... to get the best potential quality... fhg doesn't really do good VBR, in fact in mp3 lame is alone at being the best quality with VBR.... Also although lame is progressing in versions the currently recommended version of lame for qualitys sake is 3.90.2 or something like that... improvements made since have altered the balance with the --alt-presets causing needless increases in bitrate and potential problems with sound quality.. you are correct with fhg being traditionally thought of as better with transient signals, and the problems of pre-echo.. with --alt-presets tho lame has very few problems with problem samples, it spots them and throws bits at the problem... www.hydrogenaudio.org Again for the word on all things concerned with audio encoding, in all formats Jim

quoted 68 lines -----Original Message-----

> -----Original Message----- > From: Mark Kolmar [mailto:mark@burningrome.com] > Sent: 16 April 2003 22:59 > To: IDM List > Subject: [idm] crappy MP3 encoding > > > One important factor many seem to miss is that MP3 encoders are not equal. > The best-sounding are probably LAME and Fraunhofer, and the worst probably > Blade and Xing. These all sound different from each other at the same > bitrate. > > They are also not equal to themselves. LAME is an ongoing project -- they > are up to v3.93.1. There are several Fraunhofer encoders that do not > behave exactly the same either. > > Plus, LAME and Fraunhofer both have settings to trade off between sound > quality and encoding time. Fraunhofer is relatively slow, and LAME > relatively fast. > > In terms of overall character, *very* generally, Fraunhofer can sound > artificially crisp, while LAME can soften very crisp or sharp sounds. A > web search will locate a few tests people have done, and some test cases > that encoders can have trouble with. > > If you'll forgive a bit of self-promotion: > > http://www.burningrome.com/music/sounds > > Most of the MP3s are encoded at 128kbps using one of the Fraunhofer > encoders. The choice of 128kbps is for small file size -- quicker > download by modem, and less disk space on the shared server where storage > is limited. The choice of the Fraunhofer encoder is for acceptable sound > at a relatively low bitrate. The tradeoff was longer encoding time. > > A few days ago, I posted a 76-minute live performance I did for WNUR 89.3 > FM Evanston/Chicago in September, 2001. Those MP3s are encoded at 160kbps > using LAME 3.93.1 at highest encoding quality (-q0). The files are a > reasonable size, encoding was quite fast, and the sound quality is > functionally equivalent to the original recording -- especially as > compared to the FM radio broadcast. If the idea of "ambient digital > brutality" sounds vaguely interesting, please grab the MP3s and enjoy. > > --Mark > > >>> > To: BAlbers@premiereradio.com > From: "alan flood" <guile133@hotmail.com> > Cc: idm@hyperreal.org > Subject: [idm] ATP (ae) track and crappy MP3 encoding > Message-ID: <Law10-F29lrg2nRySLp000078ff@hotmail.com> > > Speaking of encoding mp3's .......... who the hell encodes at bitrates > lower than 192?????? Soulseek and the like are filled with tracks and > albums that are encoded at 128 and 160. Pretty moronic if you ask > me......even more moronic are the people who download these low quality > tracks and actually listen to them. It should be a law that you mujst > encode at 192 or higher. I guess some people like listening to there > tracks with the bandwith noise > <<< > > > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: idm-unsubscribe@hyperreal.org > For additional commands, e-mail: idm-help@hyperreal.org >

--------------------------------------------------------------------- To unsubscribe, e-mail: idm-unsubscribe@hyperreal.org For additional commands, e-mail: idm-help@hyperreal.org

On 16 Apr 2003 at 21:58, Mark Kolmar wrote:

quoted 2 lines Plus, LAME and Fraunhofer both have settings to trade off between

> Plus, LAME and Fraunhofer both have settings to trade off between > sound quality and encoding time.

Enlighten me if I'm wrong, but I was under the impression that Fraunhofer are the original developers of the MP3 compression algorhithm (sp??) which is used by *all* MP3 encoders? Is there really a massive audible difference (at the same bitrate) between the various app's? I've been using LAME since it's regarded as being the 'best' and I've certainly been happy with the results at 192 for archiving my music; however since I encoded a file to .ogg yesterday for the first time ever, I'm truly convinced that Ogg sounds better. Henceforth, I shall ogg away... just as soon as I install a user- friendly front-end version that doesn't rely on command-lines, since I'm not nearly geeky enough :-) I * --------------------------------------------------------------------- To unsubscribe, e-mail: idm-unsubscribe@hyperreal.org For additional commands, e-mail: idm-help@hyperreal.org