MP3 Pro codec demo by Thomson to be released tomorrow!

[Indiana]

Wombat, you should definitely ask this question to Creative! I have no idea what kind of drugs their development team is taking to get noise on a digital out - but there is. Just test it yourself with a digital reciever.
This is one of the good things about the Onkyo - nearly no noise even when the volume is turned up fully, and none hearable without headphones if the digital inputs are used.
BUT with the Creative Undead! SP/DIF connected to the reciever there actually is quite a lot of static plus extra noise when you move the mouse, scroll text, HD-activity is taking place (and this noise can be actually heard at normal volume-levels). Just like you'd expect from a bad analog-out - just that Creative manages this in digital, some kind of "digital-quality static/hum (tm)"...

LiquidSnake, to my knowledge LAME and Fraunhofer codecs do process the full frequency range when the HQ-switch is checked. But then I normally don't use bitrates as low as 128Kbps.


[This message has been edited by Indiana (edited 17 June 2001).]

[impact]

Ogg what?

I'll add the higher encoding rates (192 and 256) soon...

[impact]

I'll use the frauenhofer codec again for the 192 and 256kbps, but this time I'll be using the "alternate codec (High)" setting, as it is recommended to use for files with bitrate over 160 kbps. The drawback is that it takes about 40x more time to encode the whole song at this setting...

[xortam]

Thanks impact! Will you use the same track so we can get a direct comparison? Another question for you all ... what is the relative storage requirements for the various MP3 bit rates? Does the requirement go up linearly with the bit rate?

[az]

Sincy bit rate is space per time (128 Kbit per second, for instance), yes, used space and bit rate increase linearly

AZ

[Liquid Snake]

Hey, I own the Rage CD with that song, I could do some encodes (using LAME) and toss up some freq. graphs. If I had time, that is

[High_Jumbllama]

Window Media - Controlled by the big M.
Realplayer - Annoying and intrusive player.
MP3 - High bandwith encoders require a licensing fee.
Ogg/Vorbis - Still beta.
Is there any encoder for music that doesn't have some sort of significant issue?

[az]

Ogg Vorbis is out since yesterday, AFAIK, but it only allows for up to 128 kbit...

MP+ maybe?

AZ

[Indiana]

LS, please do. I'd really like to see some graphs of a Lame mp3 directly compared to the original and the respective Fraunhofer encoding with the same song/bitrate.
Just want to see if there really is a difference that I can't hear But then maybe the newer Fraunhofer-encoders are better quality-wise than the old one I've abandoned for lame.

[impact]

Okay here it is, it is the same song (it's the last track on the matrix ost).

192kpbs fhofer


256kbps fhofer

[impact]

And it looks quite interesting. Notice how the 192kbps actually goes over 21,000 hz whereas the 256kbps cuts right at 21,000 hz, but is accurately preserving everything below 21,000 hz

Btw: where can I get lame? If you tell me, I'll try it, too.

[xortam]

Thanks impact. Looks like the 192 encoding is much brighter than the 256 rate (look at the sound levels at 16-17k and above and compare with the original). Disappointing drop-offs on both at the high end. I would think this would really effect the imaging on a high-end system. Again, I'll have to try it and let my ears be the judge.

[Liquid Snake]

Guess impact beat me to it. I didn't have time to run freq. analysis. You can find LAME at http://www.mp3dev.org and encoding/ripping tips at http://www.r3mix.net

[Indiana]

Impact, you can find the latest Win compile of LAME at http://mitiok.cjb.net
Also a lot of good info about mp3-encoding can be found at http://www.r3mix.net , especially the "quality" and "analysis" sections contain lots of interesting frequency graphs in which different ecnoders are compared to the original, those are in accordance to impacts graphs: The Fraunhofer codecs don't exactly reproduce the original signal in the >16KHz area, while the newer lame encoders do.

Regards,
Joachim

P.S: An overview of the c't-test mp3 128Kbit vs 256Kbit vs CD can be found at http://www.heise.de/ct/00/06/092/default.shtml

[This message has been edited by Indiana (edited 19 June 2001).]

[Liquid Snake]

Yep, LAME reproduces the higher frequencies at 256kbps better than FhG does, it seems:

This is LAME 3.87b (I encoded this one a while back). I used plain old lame.exe -b 256 -h -m s to encode this one. This comes from the Rage CD, not the Matrix Soundtrack, but I presume the actual song was not modified for the Matrix soundtrack. If it was, my graph's invalid

[This message has been edited by Liquid Snake (edited 19 June 2001).]