Help - Search - Members - Calendar
Full Version: Extreme Artifacts
Hydrogenaudio Forums > Lossy Audio Compression > MP3 > MP3 - Tech
stegre
Hi. Anyone know if this is a known issue?
kjempen
Hmm... I don't know what settings were used in that sample provided on the web page you linked to, but when I tried encoding the sample (w/current version (Lame 3.96), setting: --preset cbr 64), I got a very different-sounding MP3. A much better sounding one than "london.3.95.1.mp3". Fairly close to the original I'd say.

And I agree, the "london.3.95.1" mp3 sounded horrible!

Either this problem has been fixed in the current version, or the person encoding that sample used a terrible set of commands/switches
shadowking
No problem with 'normal encoding' (3.96 APS)

Why is your london mp3 64kbps, 22500khz ???


I don't get your point, sorry.
shadowking
Encspot shows: lame 3.95, mpeg 2 layer III?, unwise settings = Y
stegre
Oh, maybe I should have linked to the top of the page, there's some more info there; in any event its:

LAME v3.95.1 stable 2004-01-14 (ICL7.1 compile by John33).
- All command line options at default values (i.e. no options were specified)

It happens with 3.94 too, but not 3.96 I believe
robert
lame 3.94 upto 3.95.1 should not be used. they have several bugs which were corrected now in 3.96
john33
QUOTE(robert @ Jul 7 2004, 09:35 AM)
lame 3.94 upto 3.95.1 should not be used. they have several bugs which were corrected now in 3.96

I'll remove 3.95.1 from Rarewares. wink.gif
FireStarter
Regarding 3.95.1, what artifacts are we talking about here.?
stegre
The artifacts shown here (additional details at the top of the same page).
FireStarter
hmm, sorry about that, didn`t see the link in the original post.
is this strictly for cbr?, i never have heard artifacts with my 95.1 vbr encodings.
stegre
Not sure about VBR/CBR, I haven't tested a lot of cases... worth noting, I suppose, is that it MAY be related to the 22050 samplerate of the original wav, which is the only thing even remotely unusual about the file (though "unusual" is really too strong a term). I'd feel better about encodings I've made in the past if someone did confirm "this can't and won't happen at 44100" - but I don't know that to be the case either.
naturfreak
I did a small test with this file:

LAME 3.90.3 APS original 22050 Hz : Bad artifacts
LAME 3.90.3 APS resampled to 44100 Hz with SSRC before encoding : File sounds ok

LAME 3.96 APS @22050 HZ & LAME 3.96 APS resampled to 44100 Hz: Both ok

Edit:
I just realized LAME 3.90.3 doesn't support samplerates below 32000 Hz.
Test result for 3.90.3 @22050 is useless for that reason.
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.
Invision Power Board © 2001-2008 Invision Power Services, Inc.