Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: aoTuV beta 2 + QKTune beta 3.2 (Read 13822 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

aoTuV beta 2 + QKTune beta 3.2

Reply #25 –
do you have the link to qk32 (whatever your latest is..) sources ?

aoTuV beta 2 + QKTune beta 3.2

Reply #26 –
Quote
do you have the link to qk32 (whatever your latest is..) sources ?

I just uploaded them.

http://www.rarewares.org/quantumknot/qkt32-src.tar.gz

It just contains the modified files.  Just replace the 1.0.1 versions with these (specifically info.c, psych_44.c, and psy.c).

aoTuV beta 2 + QKTune beta 3.2

Reply #27 –
Quote
It would be great to have OaQk tuned up to -q7 or -q8 

Indeed.
It would be great if you could scale your pre-echo tunings in small steps, so that for example at q0 there is no pre-echo tuning at all, at q2 some more, at q 4 even some more and at q6 and above the full effect would kick in.
If that's not possible, i'd like to see the tuning not only at q2-5, but also at q6 and above.
Hope you understand what i mean.

Edit: If pre-echo tunings wouldn't inflate bit-rates, i'd be perfectly satisfied with the second solution, but since there are also other factors that decrease sound quality at lower bit-rates, i think the full pre-echo tunings are not necessarily needed there in favor of saving some bits.

aoTuV beta 2 + QKTune beta 3.2

Reply #28 –
It has been confirmed by two people that a combination of short block tunings and aoTuV increases HF hiss.  So we can conclude that this combination (and a combo of GT3b2) is a no-goer.

aoTuV beta 2 + QKTune beta 3.2

Reply #29 –
Quote
EDIT:  Since I've found the source, I might as well make a statically linked linux binary

http://www.rarewares.org/quantumknot/oggencaqk2.gz

not quite static enough. not everyone has glibc2.3:

$ ./oggencaqk2
./oggencaqk2: /lib/libc.so.6: version `GLIBC_2.3' not found (required by ./oggencaqk2)

aoTuV beta 2 + QKTune beta 3.2

Reply #30 –
hi, guys! can anybody make ogg vorbis plugin for nero with these enhancements? something like oggdrop.
thematrixhasyou

aoTuV beta 2 + QKTune beta 3.2

Reply #31 –
Quote
It has been confirmed by two people that a combination of short block tunings and aoTuV increases HF hiss.  So we can conclude that this combination (and a combo of GT3b2) is a no-goer.

Does this apply to aoTuV + QK or only to oaTuV + Gt3b2 ?

I didn't find the time to do more tests of oaTuV beta 2 + QK beta 3.2, but my last test was really positive      So, what about QK tuned for -q6 -q7 -q8 ? so it could be the new recomended encoder at all levels

aoTuV beta 2 + QKTune beta 3.2

Reply #32 –
Quote
Does this apply to aoTuV + QK or only to oaTuV + Gt3b2 ?

Mostly the first though I suspect it equally applies to the second as well.  A new method must be devised to improve pre-echo handling without the bitrate increase as well as the introduced HF hiss. 

aoTuV beta 2 + QKTune beta 3.2

Reply #33 –
Quote
Mostly the first though I suspect it equally applies to the second as well. A new method must be devised to improve pre-echo handling without the bitrate increase as well as the introduced HF hiss.


Is there an understanding yet on a technical basis as to why short blocks increase HF noise? not really sure just plainly guessing from what I remember from skimming to the source, but does it have something to do with quantization or maybe some sort of noise companding or something of that nature? What are your ideas on how this could be improved?
budding I.T professional

aoTuV beta 2 + QKTune beta 3.2

Reply #34 –
Quote
Is there an understanding yet on a technical basis as to why short blocks increase HF noise? not really sure just plainly guessing from what I remember from skimming to the source, but does it have something to do with quantization or maybe some sort of noise companding or something of that nature? What are your ideas on how this could be improved?

That I'm not sure about yet, since I haven't delved into the changes made in aoTuV b2.  My guess is that the aggressive impulse block tunings I've been using (similar to what Garf used in GT3b1), are having a negative effect on the noise suppression of aoTuV.  Of course, that is just a 'black box' observation I'm making and that is all I can offer at this time.  Plus I'm basing this on comments from listening tests done by other people.  I haven't listened for myself where this HF hiss is appearing.  If it is appearing in sharp attacks like cymbals and hi-hats, then perhaps it is the point stereo problem which tends to boost these sorts of sounds.

I do have some ideas on how to counter this point stereo issue in perhaps a more adaptive and dynamic way (to avoid the excessive suppression in QKTune beta 3.2) but nothing has been formulated at the moment and when I do try it, it probably won't work.

For the transient-coding-without-bitrate-increase problem, there were a few ideas from Sebastian on the Vorbis-dev mailing list about sharing floor curves and noise shaping in the frequency domain.  It branched off from my 'castanets encoded with iTunes at ABR 128 kbps' comment

aoTuV beta 2 + QKTune beta 3.2

Reply #35 –
Quote
If it is appearing in sharp attacks like cymbals and hi-hats, then perhaps it is the point stereo problem which tends to boost these sorts of sounds.


yes, I haven't performed to many tests myself, but applaud tends to be more noticeable one. I never really took into into consideration as being bothersome before until I heard that sample and some metal tracks with cymbals jumping all over the place.


Quote
but nothing has been formulated at the moment and when I do try it, it probably won't work.


Hey anything is worth a shot ;-D.

Quote
there were a few ideas from Sebastian on the Vorbis-dev mailing list about sharing floor curves and noise shaping in the frequency domain


I skimmed over the paper the ideas were pretty interesting.  Thanks for reply.
budding I.T professional

aoTuV beta 2 + QKTune beta 3.2

Reply #36 –
Quote
It has been confirmed by two people that a combination of short block tunings and aoTuV increases HF hiss.  So we can conclude that this combination (and a combo of GT3b2) is a no-goer.

hum hum ... and what about megamix ? it seems to be working 
I'm confused.
 

aoTuV beta 2 + QKTune beta 3.2

Reply #37 –
Quote
Quote
It has been confirmed by two people that a combination of short block tunings and aoTuV increases HF hiss.  So we can conclude that this combination (and a combo of GT3b2) is a no-goer.

hum hum ... and what about megamix ? it seems to be working 
I'm confused.

Yeah, there are some conflicting results here.  I'm confused as well.  My guess is that the HF hiss in these combo coders (eg. megamix) aren't as bad as in the CVS code.