ok... i know all about the --alt-presets, but i see people talking about all these different switches and stuff... so what is the *best* setting for 192k cbr?

all opinions welcome

thanks

--alt-preset cbr 192 or
--alt-preset cbr 192 -Y

I don't think the -Y switch does anything with non-VBR modes: a quick test comparing the two commandlines gave same-size files.

So I say the absolute best CBR 192 line would be --alt-preset cbr 192 --lowpass XX, where XX is the lowest lowpass that you can use without noticing a loss of high frequencies. The idea is that you throw out as much high-frequency information as possible, so the frequencies you can hear will get more bits from the encoder. So for me, that would be "--lowpass 15.5". For you, it might be different.

To find your optimal lowpass, check out KikeG's thread How high can you hear (with music & lowpass).

Different file-sizes on two CBR files, same bitrate, same source?

Depends on the LAME version.

If you are using the recommended 3.90.2 version of LAME, you can use --alt-preset cbr 192. However, when LAME 3.94 finally comes out, you'll have access to newly created switches that should increase quality a little bit more.

So if you have an "encoding project" queued up, I would wait for LAME 3.94 to be released.

Don't play with it. Just use --alt-preset cbr 192 using LAME 3.90.2 or 3.91.

It's the setting I'm using at the moment to fit about 8 audio CDs on a single MP3 CD.

You'll be happy with the results, I'm sure.

Why use 192k CBR, when --alt-preset standard averages to around 192kbps on most music (similar file sizes), and is guaranteed to sound better?

A quick glance at the topic:

What is the absolute *BEST* 192K CBR LAME setting?

I took the same wave source and used dreamliner77's two commandlines to generate two MP3's. The files were the exact same size, which leads me to believe that the -Y switch had no effect. IIRC, -Y won't affect CBR, nor will it affect ABR.

Be honest with me: was I not clear with what I said? I'm questioning my own ability to communicate right now... :alien:

hello,

anybody knows what commandline switches equal "--alt-preset cbr 192"? just want to know whats behind this preset to compare it to some older lame encodes with custom 192k switches.

thx

I think he's questioning why you believe that making any changes at all to the command line of a CBR encode would change the file size.

The resulting file will be 192kbps no matter what, no?

--alt-preset cbr 192 using LAME 3.90.2 by Dibrom

Don't tell anyone I told you this but....

it's just r3mix with -b 128.

j/k. It's not just a bunch of switches, there's code-level tweaks as well.

The standard answer (the one above this post) is that it also includes code-level tweaks, and so is not directly comparable.

Excluding the code level tweaks, I think it's most analogous to:

--nspsytune -V2 -mj -h --lowpass 19.5 -b128 --nssafejoint --athtype 4 --ns-sfb21 3

compare this to r3mix:

--nspsytune --vbr-mtrh -V1 -mj -h -b96 --lowpass 19.5 --athtype 3 --ns-sfb21 2 -Z --scale 0.98 -X0

But you'd really have to ask Dibrom. Again, it's not really fair to consider them along the same lines as other command-line tweaks.

If it helps, the alt-presets are considered to be entirely superior to r3mix (it's true, read the History at www.mp3dev.org), and thus likely superior to what you'd come up with on your own unless you really know your stuff.

EDIT: whoops, forgot to edit this a while back. Once again, I have been corrected. Check two posts down.

Argh, I'm so stoopid! You can see how often I use CBR encoding.

I went back and actually created a .sfv file with both MP3's in it, and, sure enough, they generated the same CRC value. So the -Y switch does not have an effect on the output of a CBR MP3. But my initial reason for believing the files to be identical (same filesize) was obviously faulty.

Four out of my five posts yesterday were totally boneheaded. I'll bet the fifth one would have been, too, if it wasn't so short.

This has been said million times but here goes again for the sake of consistency:
--alt-preset cbr [bitrate] and --alt-preset [bitrate] don't use any code level tweaks. Only standard,extreme and insane alt-presets use code level tweaks.

--alt-preset cbr 192 in Lame 3.90.2 - 3.93.1 is an alias for:
-b192 -h --nspsytune -mj --nsmsfix 1.7 --ns-bass -2 --athtype 2 --lowpass 19.5 --scale 0.97
(only nsmsfix value can't be seen easily with --verbose switch, everything else here can..)

-Y switch is not meant for cbr/abr at all. Unless one wants to wait for Lame 3.94 which definitely is capable for better cbr 192, I suggest one uses Lame 3.90.2 --alt-preset cbr 192 for cbr192 needs.

Eeeee... The question was 'anybody knows what commandline switches equal "--alt-preset cbr 192"?'? Right?
But it's cbr preset, and youtr line is vbr (V2) with minimum bitrate of 128kbps.
And AFAIK cbr presets doesn't contain any code level tweaks.

Edit: I was writing this for over 9 minutes! Too long I see

Ok, I know I can't use the quoting system but anyway...

JohnV said that (and I'm most grateful for this!!!)...

[--alt-preset cbr 192 in Lame 3.90.2 - 3.93.1 is an alias for:
-b192 -h --nspsytune -mj --nsmsfix 1.7 --ns-bass -2 --athtype 2 --lowpass 19.5 --scale 0.97
(only nsmsfix value can't be seen easily with --verbose switch, everything else here can..)]

while I read in another quote from Delerium (from:
http://www.hydrogenaudio.org/forums/index....3076&hl=nsmsfix
)...

[Originally posted by westgroveg
I wanted to know what lowpass is used for standard/extreme/insane.
standard lowpasses at 19 kHz, extreme at 19.5 kHz, and insane at 20.5 kHz. ]

Now I want to know WHY the lowpass in --alt-preset cbr 192 is set to 19.5, when in --alt-preset standard, which is allowed 320 kbps frames, the lowpass is only set to 19. Seems strange to me... I'd go for a lower lowpass for --alt-preset cbr 192, probably below 18.

More questions regarding --alt-preset cbr 192:

- Should scaling really be put into the presets?
- what EXACTLY does the combo --nspsytune --nsmsfix 1.7 --ns-bass -2 achieve in terms of sound quality improvement?

RD.

19.5 for cbr-192 is a bit high also in my opinion. I'd go for 18-19.

I think for educated users, who can use certain tools, the scaling is not necessary, but for majority of people I think it's good for basic clipping reduction.

--nspsytune uses psyco-acoustic model with tunings originally by Naoki Shibata (which is used by all --alt-presets).
--nsmsfix adjusts masking of the mid/side frames and that way also affects ms/lr switching. (Works only with nspsytune)
--ns-bass -2 gives a bit more resolution for the low frequencies. (Works only with nspsytune)

With lame 3.94alpha 11 and further, some things will probably be adjusted for better quality for cbr-192:
--shortthreshold 3.5,15 or 4.5,15 or somewhere near.
-X 1,3 or 2,3 or 3,3 (anybody tested 2,3 or 3,3 with this? 1,3 should be safe, but 2,3 or 3,3 will probably give different, either worse or better quality than 1,3)
-Z1 (noise shaping type 1. I personally don't trust type 2 much at all.)
--lowpass (maybe 18 which gives transition band of 17960 Hz - 18494 Hz. This transition band is given by lowpass values between 17.8-18.4)
--nsmsfix (ideal value should be searched)

lol, thanks for all the quick replies... and i'm seeing all those crazy switches that everyone talks about

one other question i had... is it a good idea or not to use the -p (crc checksum) along with the alt preset?

--alt-preset cbr 192 -p

i know the crc uses 16 bits per frame which would otherwise be used for encoding, but is that a good thing to do or no??

thanks all B)

No. It wastes bits otherwhise used for encoding, as you said. It's just error detection, not error correction. If you want to make sure that the file hasn't changed, use some program to create checksums (e.g. MD5). If you use CD-Rs to archive mp3 files it might be a good idea to write each CD twice (maybe with different brands), or put the files to an archive (e.g. RAR) with error recovery information.

-edited: some additional details

I remember testing -X 2,3 in the past and the results were unfavorable. Of course, I forget if I was using CBR/ABR/VBR or the bitrate so maybe things would be different with 192 CBR (which is something I never use).

-X 3,3 seems good with ABR and VBR but it can produce bad sound with CBR. Try -X 3,3 with 128 CBR. Definitely not good (compared with -X 1,3). At 192kbps, things might be different though. Usually -X 1,3 is safe with just about anything.

Interesting thread.

Thanks Dr. TaaDow for asking the question!

just had post written up and hit reload accidently....grrrr.

Some questions, if I may....

1) Is q2 really better than q0? That seems like a bug.

2) I notice people skipping versions 3.92-3 in their recommendations...how come?

3) where can I find a website/page explaining the switches (of lame) to me. Not just
what they "are" -- I can see that in the usage.txt file, but the ATH stuff? What's that?

The psy stuff is also fairly confusing....and (this one I feel like I should know) what
is emphaisis/pre-emphasis (-e switch?)?

Thanks!
:-)

-q 0 is an experimental mode which has some quality issues, while -q 2 is well tested and reliable. I don't know if this still holds true for the upcoming LAME 3.94 though - I doubt it.




There's no real reason not to use LAME 3.92 (or LAME 3.93.1), actually - the only reason why Dibrom's compile of LAME 3.90.2 is recommended is that this is the most intensively tested compile available, which definitely works as it should. The other LAME versions are compiled with different compiler options and can thus produce different results from 3.90.2 - in at least one case, this difference has been proven to be audible, if I remember correctly (Dibrom?).

Edit: I forgot to add that, at least on my system, Dibrom's compile is also faster than the rest.

speed isn't important to me, i'm a very patient person... i'm all about quality

if need be, i would rather wait on a slower compile/encoder if it was putting out better quality than a faster compile/encoder B)