Stereo Tool

BETA057SLOW: Natural Dynamics test version

Ok, I know the Natural Dynamics filter in this beta is EXTREMELY slow; if I enable everything even with my quad core I get some hiccups. (So Brian, just don't try it).

What I want to know is if the SOUND is going in the right direction (I think it is).

As far as I can hear:
- The smoothness is good
- The weird sounds are mostly gone
- There's little (unwanted) interaction between different sounds

Making this the best version so far. But it needs 18 (!) bands.

Anyway, I first need to know WHAT I need to make, then if it works I'll check if I can improve the performance.

From the previous beta (still valid):
The CPU load is also pretty high because I'm calculating 32 different tilt levels to find the best match (so automatic tilt detection works, but it needs serious optimization).

New slider: "Declipping margin" - if you set this to 75%, everything above 75% of the maximum level is considered to be potentially clipped.

Changed slider: "Never consider samples to be clipped if they are lower than:" - rule out soft sounds, whatever they look like, they are not clipped.


Winamp DSP plugin: http://www.stereotool.com/download/dsp_ ... 57SLOW.exe
Stand alone version: http://www.stereotool.com/download/ster ... 57SLOW.exe
VST version: http://www.stereotool.com/download/vst_ ... 57SLOW.dll
VST version (No SSE2): http://www.stereotool.com/download/vst_ ... 57SLOW.dll
Command line version: http://www.stereotool.com/download/ster ... 57SLOW.exe
Linux command line version: http://www.stereotool.com/download/ster ... 20-057SLOW NOT AVAILABLE
Linux GUI version: http://www.stereotool.com/download/ster ... 20-057SLOW NOT AVAILABLE

TODO:
- Fix loading changed multiband frequencies
- Add buffer and filter for SCA output (SCA1 ok, SCA2 ok)
- Finish AGC improvement - make mono value configurable (replace checkbox by slider) 1 hour --> NO, not needed - anything else needed? -> NO
- Fix Punch
- Check what to do with new filters (such as bass AGC) - keep them, remove them, change them? --> KEEP
- Save new BASS_AGC setting in VST mode
- Loudness: Annoying cracking sound in bass. Slightly present in 5.00, worse in 6.00, maybe even worse in 6.10. Only when bass is too loud. Much worse than in Final Limiter (at same input level!) - so this clearly indicates a bug. Most likely cause: The filter that was added to remove bass artifacts....... - No, it's the louder bass. But it can be fixed by changing some settings. Default settings updated, and behavior for 'not Very strict' improved. Also Deep bass boost and Very deep bass protection are enabled for latency 512 now.
- Fix crash at program close
- Fix VST plugin version (does not run)
- Dynamically drop 'Allow louder highs, even if it causes vibrations' to 0 when bass filter suspects noticeable voice vibrations. 1-2 hours
- Reduce Loudness CPU load days?
- Check and remove static variables
- Finish new de-essing filter (check what to do with the settings, remove at least some!) 1 day
- Convert Multiband input to MONO, then use arrays [2][4096] --> should give speedup. - FAILED
- Natural Dynamics: Fix or remove transient boost
- Natural Dynamics: Add expected + strength slider per band
- Finish declipping filter (clipping level detection + level reduction in dB). 1. Figure out why removal of unwanted frequencies causes flat lines at high quality setting with small overlap. This causes distortion, with this fixed repairs are MUCH better. 2. Fix MP3 correction, automatically scale down when this deteriorates the sound. --> TOP part fails!
- Declipping filter: Fix low latency behavior
- Always oversample clipping (configurable)
- Declipping filter: Change detection at lower input levels.
- "Test Right Channel" in "FM Transmitter Calibration" does not work since v6.10 (standalone) (bojcha)
- Declipper window close function - check! - Seems ok
- Scopes black background?
- Add AGC start level
- RDS issue reported here: viewtopic.php?f=15&t=3703&p=11524#p11524 partially solved
- Natural Dynamics: Voices, especially in chorus, still sound weird.
- Natural Dynamics: Smooth out different bands more to reduce low-bitrate-MP3-like sounds.
- Natural Dynamics: Optimize situation with no interaction between bands for performance.
- Declipping filter: Optimize tilt detection for performance
- Declipping filter: Optimize peak matching for performance
- Update presets? (BASS_AGC etc.) 1 day
- Finish blind interface
- Channels L/R swap in stand alone version when changing filtering/quality (?) (eldoradofm)
- Move pre-emphasis to end of processing 1 day
- Save all new settings, also through VST interface
- Change version number 1 hour
- Release 1 hour
- Add lowpass filter for stereo signal (will cause a lot of extra latency!) - it might be possible to avoid this latency using a Hilbert transform
- Add smarter clipping detection. Maybe something much simpler suffices: Current clipping detection with threshold + flat line detection
- Declipping filter: Automatically override the 3 clipping level sliders if the clipping level is detected very clearly (clear thin spike in sample value histogram). - NONSENSE, this is already done by the histogram function. But it can be made a BIT better - I think - by automatically LOWERING the 'always clipped' slider if a lot of data is present at the highest few bins (but care is needed for DBN - Jack is Back like tracks)

What effect smoothing setting do you recommend?

My actual setting:


I myself always try to get a smooth boost on voices and bass to sound more dynamic, under no circumstances may it boost highs too much, that causes a more scratchy sound. And i think i got it with this config.

Trying it now. Little CPU increase noticed on i7 860. Maybe 2% more? (But it was only using 6% before with everything but FM enabled, so that's a pretty big relative jump I guess)

So far pretty amazing sound Hans. I do have one suggestion with Natural Dynamics... to allow for the "Effect Smoothing" to be independent for bass, mids, and treble. I like what it's doing on the lows and mids, but on the highs I would like to use little to none, because it improves the crispness of the transients a lot.

[edit]
or maybe a button that disables that from the highs, like the top 3-4 virtual-bands (or top 7-9 actual bands), if that's easier
[/edit]

@Jesse: I already planned to add that

So, only a 2% increase! Wow! I'll have to try it on my new i7 laptop (which is probably a lot faster than my 2 year old development system).

Anyway, this means that you agree that this version sounds much better than before, right? What I'm doing here is:
- 18 bands
- Some smoothing between bands
The smoothing between bands reduces weird sounds, but gives too much linking between frequencies, which was solved by doubling the number of bands.

What I want to try is if I can think of a way to process each band separately, without the huge (imo) CPU load. If I can't I'll definitely include something based on this latest version (maybe with even more bands!). Ideally I would find the loud frequencies, and dynamically create a band around it - but that would REALLY cause the CPU load to rise way to high, even on your system. (Unless I can find an efficient way of doing it).

Every possible option on and all settings on high on my i7 laptop about 13 to 14%

Need to check the sound one of these days, alot to catch up it seems.

Natural Boost ON Only - Beta056 vs bet057 - CPU Usage on old AMD X2 5600+

Beta056 - 20%
Beta057 - 40%

@Jesse - Do you have new 'undo'? If yes, can i send you one track to porcess it with 'undo' only ?

I did that with GPU processing, and then manipulated in 3D space, IO of FFT of course. Totally non-conventional and something that wouldn't run on 99% of systems without decent CUDA cajonies. Also, it wasn't that great of an outcome, my "old" mid-side multiband expander got lot better results, and way easier to use. Maybe if I hit an epiphany to the processing itself, it could work better, but yeah... scrapped it for now.

I don't actually. Been trying to leave Leif alone so he can stay focused. We tend to like to shoot the "crap" a LOT when we do.

---

To clarify my earlier comment on CPU, I'll come clean on the i7 860... I finally decided to overclock it. So instead of 2.8gHz, it's now a top normal speed of 3.19gHz (133mHz FSB base clock turned up to 150mHz). Plus it will "turbo boost" to 3.6gHz. Memory goes along for the ride from 1800mHz to 2030mHz. Thankfully my ASRock board supports "untied" clocking, so my PCI bus stays the same. (But I'm also overclocking my EVGA GTX460 "Free Performance Boost" edition, to the "FTW" edition clock speeds, using EVGA's own OC tool)

So it's performing more like a top end i7 setup would. My mainboard will actually accept up to 2600mHz memory though. Probably not worth it to upgrade when the price comes down, since I'll maybe be onto a faster overall system by then anyways. And 2030mHz ain't too shabby.

@JesseG: Those are really impressive numbers... I have a Q9450 quad core CPU running at 2.66 GHz which shouldn't be that bad, but my memory is running at only 333 MHz (The main board can handly upto 400 MHz, but only if I put at most 4 GB in it. And I have 8).

I've been looking at a lot of tilt measurement data and I've tested some optimizations on paper, and based on this I think I can lower the number of steps to determine the tilt from the current 32 to 11 without loosing any precision. So as soon as I have some time I'll try to implement this idea and see if it really works.

@Hans, can we also expect a tilt correction slider for the output, with a 60 Hz square wave test tone for adjusting it?