Well, just couldn't wait to listen to original CD versus "--r3mix" LAME encoded mp3. Interesting, very intersting!

I like listening to music late at night; no it's not my imagination, the power supply is a bit cleaner!

Not the highest of hi-fi that I was using for this quick test. A CyberHome DVD player, so I can compare the mp3s-off-a-CD and the original CD from the same machine. Ok, maybe not the most scientific test, but from some cool tunes I've just got hold of (some Ambient Lounge of all things) there's a distinct audible difference between mp3 and original track.

To my ears, the difference seems to be mostly in the imaging. From the original track, there was a very clear "source" that I could point at. The overall "sound" of the mp3 track was the same; the glaring difference, was that I wasn't able to hear this same "source".

But I don't want to detract from it too much, this LAME encoder is very, very good now.

So... suggestions for changing the encoder settings to something more transparent? I'm new at fiddling around with these subtle settings, so any guidance is appreciated!

I'll persuade my girlfriend to help me do some double-blind listening tests tomorrow night; I'll prepare some different samples and see what happens. Got to buy some more speaker cables tomorrow so I can rig my "proper" hi-fi, not just this spare kit that I was using tonight.

How much do you like the "real thing", or are you prepared to accept a substitute? I still quite like butter, actually

eFrisky:

Give --dm-preset standard a try. It has been shown numerous times to provide higher quality than that of --r3mix. You may find this difference is enough for you.

As for double blind listening tests, you can perform them yourself with a program you can find at http://www.pcabx.com

There is actually an abx program that Sphoid and I are working on as well, but it is not yet ready to be released.

I believe
"--dm-preset standard" is close to transparent on the medium range hi-fi systems.

You might to try out --dm-preset insane (as far as I know, it is still being tweaked).

Or as Trnsz mentioned 100% close to transparent
"-insane -bw 21000 -tmn 36 -nmt 20 -scale 0.96"
I have no idea about most of the switches, maybe someone is willing to explain.

Just want to mention, since it's probably not obvious for everybody that --dm-preset switches are for Lame encoder while Trnsz's switches are for MPC (mppenc).

Ah...

I thought those switches are some super super secrets. I was scratching my head trying to figure it out. Now it makes me feels much better.

I might be wrong about this, but rather than using --dm-preset xtreme, you might want to try using --dm-preset standard -Z until Dibrom has time to rework the insane preset.

What are some advantages of using -Z and without -Z? I am still quite unclear on the -z manipulation.

Thanks for all your help and advice guys, I really appreciate it

I would rather not people use this switch or get used to this switch. At the moment, yes, -Z can fix some problems --dm-preset standard has on bass pulse samples, but I have already addressed them in my new custom compiles ([b]without having to disable noise shaping 2 and it works in ALL dm-presets!) which will be out very shortly. So sit tight, and don't worry about it

Does this mean to expect some CVS commits in the next few days? :-)

Of course

Once I get some of this stuff a little more finalized all the --dm-presets will be updated and cvs commits will be sent accordingly.