version 1.05e

---- DOWNLOAD LOCATIONS REMOVED, IMPROVED 1.05f available ----


changes:

- New mppenc, version 1.05e

* --insane do not a full bandwidth encoding anymore

* --quality x.x with x.x between 0.0 and 9.9 exists.
--quality 4 is radio, 5 is standard, 6 is extreme.

* bitrate dropped again

* I expect some quality problems now, find out where

* Currently there are a lot of situations where mppenc wastes a lot of
bits for nearly nothing. At low bitrates this is very easily audible,
the quality changes from sample to sample. Soem sound with --thumb horrible,
other quite nice (without the reference). Goal is to make the quality
more constant and reinvest the saved bits later to achieve a better and
more constant quality for the same bitrate.

I get an (Internal Error) when I use --quality 10. --quality 9.99 works as expected, but --quality 10 doesn't work. Using castanets.wav

"--quality x.x with x.x between 0.0 and 9.9 exists"

Two-digit quality switch is not implemented... yet? :diabolic:


edit: damn you, spase

I get internal errors too, when using --quality 10, however the resulting file does play in Winamp. The 0.93 Winamp plugin wouldn't playback files of q < 5. Version 0.94 seems to have no problems.

Perhaps <5 use the "petri net simulation"

ahem.. pns

/Erik

What sorts of quality problems do you think there would be? Is it likely that only q values less than 5 would have such problems?

--quality 0...7

Problems with HF precision, ringing, stereo imaging and much much more.
Problem with very silent music.
--
Frank Klemm

Any particular reason why -q 5 ("standard") is now possibly non-transparent where it might have been before?

Good question. That kinda worries me. :-)

uh-oh, this is starting to be really scary. guess it'll take some time until I seriously start encoding mpcs with versions >1.03...

mppenc-----Chemical Brothers------Richard Thimpson
--------------"Block Rockin' Beats"---"Batsheba Smiles"
v.------------------kbps-------------------------kbps
1.02---------------183--------------------------200
1.04---------------150--------------------------195
1.05e--------------136--------------------------185

--xtreme (--xtreme switch still works in 1.05e)

Time for some listening tests.

Note that 1.05e is not a public release. It's primarily for testing and improving.

One question:

If I were to leave everything unchanged, like just leave everything default would I get a "standard" switch in previous MusePack Encoders?

Cheers
AgentMil

I guess the last safe encoder is the 1.04 version up till now. Tests proved that the quality was the same or even better on the same bitrate.

This 1.05e can be considered as a testcase for some new code??

I'm keeping my hands off this one for the time being, btw thanks for explicitely saying that quality issues are expected. It keeps the speculating down.

Yes, the default is and always will be standard.

On the other hand: the --quality parameter is not restricted to 1 decimal.
I was trying some --quality values in combination with --verbose to get an idea what Frank had put under the --quality switch (Just curious:o).
I noticed that -q 6.5 gave --nmt 8.9 while -q 6.6 went to --nmt 9.1. Indeed with -q 6.65 --nmt showed up as 9.0. (As did --quality 6.66 :diabolic: ).

To skip that discussion, I'll be the first to admid that this won't make an audible difference But hey, it gives the tweakers a hundred times more options.

I know that 1.05 is considered beta (or even alpha) but I found that besides giving
-q >= 10 it is also possible to give negative values. Something that can easily happen when dealing with command line options, as it happened to me . This will be simple to fix with an input argument check though.
Also the switches --nmt and --tmn are still present in this version so this version might be the one everyone likes .

From the --longhelp it showed that a new (set of) ATH's was introduced and defaulted:
--ltq 10...20: Other ATH with different HF roll-off (dflt)
(Does that account for the decreased bitrates ??)
Well I guess it's in everyones interest to help Frank and do some serious listening tests with this new stuff.

Frank: I'd like to suggest that the option -q is added as an alias for --quality.
BTW. I realy like the improved scalability.
--
Ge Someone

edit: I hate the fact that preview shows the message completely different