On Sat, 08 May 2004 06:36:04 -0500
<[EMAIL PROTECTED]> wrote:
> This is the way i went:
> i made comparisons on three computers with diferent sound cards,
> i ripped hundreds diferent samples from 30 cd's (old cd's, new cd's
> plus diferent genres and styles) (my cd collection has 455 cd's),
> i encoded the same file as many time as nedded.
> Every time i encoded one file, i used diferent options
> and i made comparisons listening carefully on every computer,
> i was picking up usefull options and dropping those i didn't like
> finally i started to work with values and diferent positions for the
> options.
So you used "ordinary" music. Typically you do such listening tests with
known bad samples.
> Thats how i got this:
> lame -q0 -X6 -ms -d --lowpass 20.05 --cwlimit 11.025 -Z --scale 0.98
> --scale-r 1.1 --scale-l 1.1 --ns-sfb21 2 --nspsytune -v -V0 --vbr-rh --abr
> 172
>
> I know that the "--scale" equals to "--scale-r" + "--scale-l" but they do
> separate left and right also while usin "-d" and "-ms" (it didn't work the
> same way with only "--scale")
> "--scale" might be for "-Z" and "--ns-sfb21". They all work better together.
No, sorry, this isn't the way it works. Feel free to have a look at the
source code in libmp3lame/lame.c. If you search for "scale_left" you
will find a location with 3 "for"-loops. The first one does "--scale".
It's followed by the second one which does "--scale-l". The third one is
"--scale-r". So with your options both channels get first scaled down to
0.98 times the original value. Then the already scaled down left channel
gets scaled up to 1.1 time the previous value which results in a total
scale of 1.078 times the original value. Then the same happens with the
right channel.
It may be the case that in some rare circumstances the output is a
little bit different with this two-step scaling, but this can only be
minor rounding errors. I can't believe those rounding errors have a
hearable impact on the output.
> Good reasons to add this settings as a preset are:
> produce high fidelity stereo sound,
> produce small files compared to the quality of the sound,
> produce ABR that can be used for streaming instead of VBR ( vbr sometimes
> too big)
> with ABR, the files don't get to big when the encoder decides to use more
> bitrate,
> "--abr 172" is the best for the ATH while working with ABR (i worked around
> with <160 and >192 and didn't like it)
> This is a lot better than "--preset 172"
Since only a limited number of persons (so far only you) tested these
settings, we won't add it. If a lot of people do test it with the "known
bad" samples in a double-blind-listening-test, and if they all agree
that your setting is better, we may add it.
Maybe you can ask at hydrogenaudio for a listening test of your setting.
Bye,
Alexander.
--
I'm available to get hired.
http://www.Leidinger.net Alexander @ Leidinger.net
GPG fingerprint = C518 BC70 E67F 143F BE91 3365 79E2 9C60 B006 3FE7
_______________________________________________
mp3encoder mailing list
[EMAIL PROTECTED]
http://minnie.tuhs.org/mailman/listinfo/mp3encoder
