All times are UTC+02:00




Post new topic  Reply to topic  [ 180 posts ]  Go to page Previous 17 8 9 10 1118 Next
Author Message
PostPosted: Tue May 22, 2012 7:17 pm 
Site Admin
User avatar

Joined: Mon Mar 17, 2008 1:40 am
Posts: 11425
6.30 had other probleem with bass sounds...

Anyway, I guess I can change the code to use 2048 samples for latency. This will add 1024 samples to the existing latency.
An advantage is that other artifacts will be reduced.

Btw: the distortion is probably also present @ 2048, but it stops twice as fast. So it makes sense that it would be very hard to hear.


Top
   
PostPosted: Tue May 22, 2012 7:26 pm 
User avatar

Joined: Tue Mar 17, 2009 2:56 pm
Posts: 4229
That would be awesome. I just hope that all other filters can work exactly same as with 4096, like, for example, HPF and multiband.


Top
   
PostPosted: Tue May 22, 2012 7:31 pm 

Joined: Sun May 02, 2010 11:26 pm
Posts: 547
Quote:
6.30 had other probleem with bass sounds...

Anyway, I guess I can change the code to use 2048 samples for latency. This will add 1024 samples to the existing latency.
An advantage is that other artifacts will be reduced.

Btw: the distortion is probably also present @ 2048, but it stops twice as fast. So it makes sense that it would be very hard to hear.
Does that mean 4096 will change to 4096+1024 = 5120 ?


Top
   
PostPosted: Tue May 22, 2012 8:42 pm 
Site Admin
User avatar

Joined: Mon Mar 17, 2008 1:40 am
Posts: 11425
Yes


Top
   
PostPosted: Tue May 22, 2012 9:22 pm 

Joined: Sun Dec 12, 2010 2:26 pm
Posts: 885
CPU load concern with adding 1024...

Since there's no real estate for an additional checkbox, suggestion:

Have the 512, 1024, 2048, and 4096 buttons toggleable by clicking on them again.

Example:

Click 512 once, and it is 512. Click it again, it becomes edit: 1536. Click it again, it goes back to 512.

For 4096, this would be once = 4096, twice = 5120, three times = 4096.

Rationale: Not all of us hear much of any change, good or bad, with these minor changes, but do notice CPU load increases. Without being able to toggle, users that hear no change, and thus derive no benefit from the change, will be forced to have a CPU load increase, or decrease samples to 3072.

Edit: Not sure why my brain didn't add 1024 to 512 properly... :oops:


Last edited by Brian on Wed May 23, 2012 7:26 pm, edited 1 time in total.

Top
   
PostPosted: Wed May 23, 2012 3:10 am 

Joined: Thu Oct 15, 2009 12:17 am
Posts: 17
Hans-
What did you remove from beta 009 standalone? The file size of beta 009 is 686 KB as compared to beta 008 which is 892KB. Beta 009 seems to install and work fine :o


Top
   
PostPosted: Wed May 23, 2012 10:00 am 
Site Admin
User avatar

Joined: Mon Mar 17, 2008 1:40 am
Posts: 11425
The non-SSE2 version :) - the build failed...


Top
   
PostPosted: Wed May 23, 2012 3:34 pm 
Site Admin
User avatar

Joined: Mon Mar 17, 2008 1:40 am
Posts: 11425
Ok. I'm now testing which filters are causing problems at low latency settings...

Filter order:
- Declipper GOOD (clipping restoration is a bit less effective, but that is to be expected
- Highpass IIR GOOD
- Phase rotation IIR GOOD
- FM Hiss GOOD
- Noise gate Some weird mild artifacts around 200-800 Hz, not problematic
- Stereo Image Bad and broken at 512! Fine at 1024+ - FIXED
- Stereo Boost Some soft artifacts for really low freqs, but at a limited level, and only for DC offsets in the L-R image. Acceptable for now.
- Stereo Image post
- Natural Dynamics Skipped
- AGC weird deformation during volume drops, very low freq, not really problematic. And some artifact in case of extreme bass levels - also not problematic
- ARTIFACTS CLEANUP START
- AGC bass BAD
- Multiband BAD
- Bandpass BAD
- Bass Boost BAD
- Highpass post filter
- BS412 collection N/A
- Singleband AGC style or Singleband AGC style gives weird artifacts during drops
- ILDP (obsolete)
- LDP (obsolete)
- ARTIFACTS CLEANUP FINISH
- Lowpass post filter
- Phase rotation Not used at low latencies (IIR is used instead)
- Pre-emphasis GOOD
- De-esser (obsolete) Skipped
- Final Clipper (obsolete?) Skipped
- BS412 protection Bad at lower latencies with aggressive settings or with Remove Remaining Peaks at low levels
- Loudness Weird frequency holes at low freqs - will fix that - FIXED
- Hard Limit Non-intrusive artifacts if you drive it really hard. But it was never intended for that anyway

So the biggest problems at low latencies occur (as I expected) in AGC Bass, Multiband, Bandpass and Bass Boost.
For Bandpass there's a non phase linear version to bypass the issues. AGC Bass, Multiband and Bass Boost are just bad.

Basically these are just the 4 filters that do something with bass.

Possible solutions:
- AGC Bass: Turn off, or use a far less steep slope at lower latencies
- Multiband: Redesign will have less bands and especially less bass bands; band separations will be far less steep. That will help a lot.
- Bass Boost: Hm.... Is this still useful with the new bass shaping mechanism in Loudness?


Edit: Confirmed now:
- Took a preset
- Turned off AGC Bass, Multiband, Bandpass, Bass boost
- Turned off Loudness ILDP
- Latency to 512
- FAR less artifacts than before!


Top
   
PostPosted: Wed May 23, 2012 4:19 pm 
User avatar

Joined: Wed Nov 19, 2008 7:44 pm
Posts: 1169
Location: Bulgaria
I advice no multiband number change . Because in my opinion less bands will cause more compression for more constant sound and more work for the limiting and loudness panel , which will cause more oversamples and more CPU usage.
What is needed here is gating level for the MB like i said before aaaaand ... TARGET LEVEL like in AGC ! ((ooops may be the soft limit sliders for every band is the same ?)) don`t think so because there is "maximum band level" which is not like the target level slider in AGC
think about that . If this options are fact ...we will use very accurate the clipper in MB ( it will do the best it`s job then ever ).
IF these rules are in use there will be constant level like no other . Slider (Limit...Compress) is a good one ...very good one.
Bass protection in "Limiting and loudness" panel is doin well i think ...what more ?
Just good forming of signal , good filtering and good limiting .thats it :)


Top
   
PostPosted: Wed May 23, 2012 4:24 pm 
Site Admin
User avatar

Joined: Mon Mar 17, 2008 1:40 am
Posts: 11425
@DJ-DOGGY: In fact, having less bands should give *better* peak protection and a more constant level going into the clipper (Loudness). Also, having wider bands with less steep cutoffs will allow far more agressive multiband settings without getting a bad sound.

Except for that, I'll make what you describe - I want to make an AGC-like compressor (similar to the current AGC but aimed at rapid response) and then use one of those for each band.


Top
   
Display posts from previous:  Sort by  
Post new topic  Reply to topic  [ 180 posts ]  Go to page Previous 17 8 9 10 1118 Next

All times are UTC+02:00


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
Powered by phpBB® Forum Software © phpBB Limited