Stereo Tool

Hello Hans


Listened the sound of the last version. All OK. I Proceed to verification on a transmitter. Betta 2 showed also not very much bad results but only in the mode 2 band.....Can it is needed to apply pre-emphasize for in end
de-emphasize for out ? For a transmitter can it be separately?

Soundmanfm

i think you only need to leave the agc eq-reaction and this peek removing, all other same as in v5.00.

I noticed swinging up down at the change of level on input...FOR stereo_tool_standalone_installer_BETA!!!!!!


But not on all songs!!!! On the sound of drum and bass guitar.

Soundmanfm

Can distinguish low freg in a separate band?

You mean also completely get rid of RMS smoothing?

One final update is building now where I restored the original hold time, BUT if a short peak causes the volume to drop, the hold time is much shorter. This has no effect on songs like that Celine Dion song, but it does on songs with loud bass kicks and only soft other sounds...

If nothing helps, old AGC from 5.00 with current: 'Remove Remaining peeks', EQ-Detetecting and (i forgot) gater.

Well, I've just uploaded a new version where the hold time is dynamically changed based on the input (short spike: Short hold time).
RMS Smoothing level is set to 110%, which is enough to get differences in output levels below 0.5-1 dB.

It is a lifetime eater !
or:
A mission impossible (you can improve an AGC, but it will never be really good).

Hans, it would really help if there would be more user feedback, because this AGC thing is (obvious) so complicated.

But with the naming scheme you use it is impossible !
Using "dsp_stereo_tool_installer_BETA.exe"
or "dsp_stereo_tool_installer_BETA2.exe" no one really knows what Beta he or she uses and more important, everyone will miss some of the beta builds (because you overwrite these with new "dsp_stereo_tool_installer_BETA.exe" or with "dsp_stereo_tool_installer_BETA2.exe" every other hour).

Please (I have suggested it before), use a naming scheme like this for the beta builds:

"dsp_stereo_tool_installer_v5_BETA_xyz.exe"

So maybe the last build (sorry I have not counted all builds)
could be:

"dsp_stereo_tool_installer_v5_BETA_027.exe"

Please !
Why not ?

Hans..

Celine Dion (My heart will go on)
song words: "Spaces between us" .. try with AGC and without........ Now try with AGC down-speed almost to 0

Agc works perfectly with AGC almost at 0 but then need much better "peek remove" or compressor/expander to compesate till level reach.

I've just compared the latest version against version 5.00, and - with the same settings - there is little no difference between the two. I played version 5.00 on the left channel and BETA on the right on headphones. Version 5.00 is a bit better, because in the BETA version the bass is counted less strongly (if I switch it to RMS there's no difference anymore)! Due to that, the voice is louder compared to the rest of the sound, which results in a bigger volume drop.

Question: Is this difference really noticeable? I can probably find a way to fix it and restore the 5.00 sound - if it's worth the effort. (How: Process band 1 twice: Once with RMS bass and once with reduced bass. Output the sound in the reduced bass version, but lock band 2 above the RMS based version. There's a risk though that this might increase the effect of bass on band 2; to get rid of that I would have to lock it above the maximum of the 2 band 1 volumes. Which might be difficult to configure - but should give better results than either choice separately.)


The default settings are different though: In 5.00, I used a down speed of 0.00025 and an up speed of 0.000005; in BETA that's 0.0003 and 0.000004. (BETA should sound closer to the original, 5.00 should have less volume differences in the output but more 'pumping').


AGC down speed near 0: That leaves the start of loud sounds way too loud. The thing is, if there's a loud voice you don't want it to respond, but if other loud sounds are suddenly present you do. That would mean that Stereo Tool would need to be able to distinguish voices from other sounds - and on top of that, if there's no music playing, the volume SHOULD drop on a loud voice...


Question: Is the current version in any way worse than 5.00? (When testing, please make sure that 'RMS peak smoothing' is either turned off or set at or below 110% - 110% should suffice now!)


Planned TODO's:
1. By default, turn RMS smoothing OFF
2. Limit the maximum volume boost value (now 250%, lower it to 125%!)
3. Tune 'hold time' speedup (dunno if that's really needed, results are *much* better than before already), and allow turning it on/off
4? Tune up/down speed (feedback welcome!)