Stereo Tool

Uploaded (BETA3A).

I am testing AGC ...
I have question. Peek remove, is it work in same time with AGC or .. first remove peek and then AGC detects signal with peeks removed, or AGC is cotroled by removed peek signal?

btw: bass boost not working when Loudness is used, or when second band in AGC is used (standalone)

First AGC is done, then if the remaining signal still contains loud peaks (RMS volume is calculated again) the volume is dropped more.
Found a bug in AGC 2nd channel and one in bass boost, fix coming soon.


I'm also restoring the v4.22 highpass filter (the new one in v4.23 did have a bit less artifacts, but it's not filtering very strongly, has effects upto 4 times the filtering frequency, and uses more CPU processing power. Instead I'm now going to use the v4.22 filtering with an option to filter less steeply, that should still give much better filtering characteristics without (much) extra artifacts.)
The non-phase linear highpass filter will also stay, and will remain the default selection for lower latency settings.

Update available as BETA3A.

Fixed:
- AGC / Bass Boost doesn't work sometimes -> should be fixed now
- AGC: Frequencies between 150-250 Hz could sometimes be far too loud -> fixed
- Highpass filter: Similar to v4.22 (but less steep filtering, because in v4.22 the latency was bigger). Slightly reduced the steepness further to reduce artifacts.

Note about Highpass filter: This filter has a bit more artifacts than the one I used in previous BETA's, but the big advantage is that it has a much flatter frequency response above the filter frequency. I think the artifacts and slightly less steep filtering are acceptable.


Edit: I just ran a new performance measurement, and compared to the version with the old highpass filter and dualband pre limiter, the CPU load has dropped by another 13%!

- Nice CPU load drop
- Second band in Prelimiter now works with Bass Boost and loudness


....

I get weird repetitive clicks when I set block overlap to about 20 % (well anything except all the way to the left), mostly notable in quiet parts.

But cpu load dropped alot, to 20 %, used to be around 45 % for me...

I was listening nelly furtado - say it right, it's got a really loud bassdrum, and it distorted the whole song when it was there. i get alot of distortion, and clip levels of the multiband compressor are suddenly very high (i suppose that's what the black bars are?), while the agc bar is very low (high reduction?) . Which doesn't make sense to me. Might be my settings, working on finding out...

*edit: weird, I restarted stereo tool, and the distortion is gone.. but the clicks are still there

hi Kakoon
Use overlap 0 .. much clearer sound.. if you have high CPU load (but in last betas cpu load is much lower) turn of loudness and push little Final Limiter but not too much in "hard red".

For AGC tune PreAmp so the meters are around half.

Yup, use Overlap=0

I've just (2 minutes ago) uploaded a version where I've fixed most of the bugs in Overlap>0 mode. This is just to make sure that the sound quality isn't better in Overlap>0 mode (I expect there'll be little difference, and Overlap=0 might even be better except for the highpass filter... - will test it later.)

When I created Overlap=0 mode the quality was clearly worse than the Overlap>0 quality, but I've fixed a lot of things since then.

IF you want to test Overlap>0 mode, make sure you set the Overlap value BEFORE the audio starts to play (which isn't possible in the stand alone version) - there's still a bug there that in some cases causes clicking (if that happens, restarting should help). Also, you might need to increase the AGC top left slider, because in Overlap>0 mode the AGC output is about 0.5 dB lower.

If the quality in Overlap>0 mode is NOT better (as mentioned, I'll run some tests myself tomorrow, feedback from others is of course also very welcome), I'll remove it completely (the CPU load for Overlap>0 is higher, so if it doesn't have any benefits there's no reason to keep supporting it.)

Hi Hans,
Last 3-4 days .. i had StereoTool crash 3-4 times on both FM and stream PC. About once a day.
ST is totaly freezed in that moment and i must kill process from task manager and start ST again.
I thought it was a coincidence first time, but today it happened again, 3rd time on transmitter site.

I am not sure from which beta this started... last 2 or 3 versions... maybe from first bass boost version or later.

Bojcha: Ok, I'll attach a debugger tonight to see what's wrong (usually that indicates the culprit within seconds).

First guess: I'm using a pre-allocated memory pool for processing steps. It might be that I'm claiming too much memory and using memory outside of that pool. Or that I forget to free chunks of data (that was the reason why bass boost and some other filters didn't run properly a few days ago, which was also an indication of memory problems and - at least internal - crashes. I though I had fixed it but now I guess I haven't.).