Stereo Tool

So one band reacts to the whole spectrum, and the other band to > 200 hz. So when there's alot of bass, the first band reduces volume, and the other band doesn't reduce as much because the bass is (partly) filtered out. However, when there are alot of highs, the first band also reduces volume, and the bass is reduced, right? So effectively in the second case it works as a single band AGC? Or is my theory wrong here?

Exactly! But usually the bass level is much higher than the highs level, so usually this doesn't happen. Loud bass sounds that are not constantly present occur very frequently though, so that's why I added the 2nd channel.

I just found a bug in the AGC filter that strangely enough only affects the left channel. It's sometimes processed VERY differently from the right channel. Working on it now.

Building an update now. This update also contains a fully functional quality slider! (All filters are working properly if you lower the quality setting).

Uploaded as BETA3A.

About the Quality slider (located next to "Block Overlap" which must be set to 0 for the Quality slider to function):
- Maximum = best quality.
- Center position (50%) corresponds to "Normal" quality in v4.22.
- 25% corresponds to "Low" quality in v4.22.

Note that setting this slider a bit below 100% has very little effect on the quality, but a lot on the CPU load. At low settings, the effect of a change of this slider on the quality is very big, at high settings it's very small.


Note: In v4.22, Normal and Low quality also means that certain filters are turned off. That's not the case when this new slider is used.

Uploaded another update as BETA3. Reason: Overlap>0 is no longer reachable in this version.

Changes:
- Restored quality/latency settings in MAIN window
- Removed separate quality/latency window. (That's why Overlap>0 is no longer reachable).

NO changes in processing!

I'm not sure yet if I should re-enable choosing Overlap>0 as an option, will do some tests tomorrow.

URL's:
Winamp DSP plugin: http://www.stereotool.com/download/dsp_ ... _BETA3.exe
Stand alone version: http://www.stereotool.com/download/ster ... _BETA3.exe
VST version: http://www.stereotool.com/download/vst_ ... _BETA3.dll


Edit: After running some tests with the latest BETA3A, I've noticed that in Overlap>0 mode there are more highs. And they seem to sound slightly better (although I'm not really sure about that, the difference is very small). Overlap=0 seems to suffer a bit more from low bass rumbling. On the other hand, Overlap=0 has more lows, with less artifacts. At the moment I'm not really sure which one sounds better, I'll look into it tomorrow.

(Note: This is at latency 512, at high latency settings the difference is very small).

Wow, cpu usage at low quality is 8-12 % here, never been that low before (4.22 was 40 % at low), very impressive! And I don't hear that much of a difference lowering the quality...

I do hear a big difference between 4096 and 512/1024/2048? I miss high-end sounds in 4096, which are present in the other buffer sizes.

Are you using pre-emphasis and listening to the Winamp output (de-emphasized)? I've just ran a test and it looks like the de-emphasis filter in lower latency modes isn't working as well as I had expected, leading to differences in the frequency response. When I turn that filter off (by, in the FM settings, enabling 'pre-emphasize output') there's very little difference - in fact I'm getting slightly less highs at latency 2048 compared 4096. Differences are in the order of 0.1 dB - negligible. Tested with "FM Loud Bass (Europe)" preset.

I think I may have found a way to move the benefits of Overlap>0 mode (there are some...) into Overlap=0 mode without increasing the CPU load. Will test it in the coming days.

Hi Hans,
Nice panel for CPU usage/latency..
For that, i notice that for 512samples .. second band in AGC is much better to be off. Something there forces second band in AGC to go down.

Something bother me last days from day when you fixed trumpet filter (crashing), i think it's from that version. I'll try to describe.
The sound is like i am playing a 32kHz audio files. Much more is if i use more loudness (curently is at 1.90-2.00x at transmitter). I guess that trumpet filter is connected with loudness in some way?. Most noted is when high level vocal (female most) on mid range is playing. The voice is not clean but it's like little downsampled and sounds like constant super-small distortion. Many times highs are sounds like that.
It's funny that i heard that first time 2-3 days ago in my car when i listened radio. My speakers are not so good but have nice mids. and highs. Loud "S" sounds like they are more downsampled then distorted.

Luckily i always have last 10 betas and if i have time tonight i'll try erlier one to compare.