Stereo Tool

For comparison:

Version 4.22 WITHOUT Trumpet:
Winamp DSP plugin: http://www.stereotool.com/download/dsp_ ... r_BETA.exe
Stand alone version: http://www.stereotool.com/download/ster ... r_BETA.exe
VST version: http://www.stereotool.com/download/vst_ ... l_BETA.dll

Latest version WITHOUT Trumpet (not changed in the last few days):
Winamp DSP plugin: http://www.stereotool.com/download/dsp_ ... _BETA2.exe
Stand alone version: http://www.stereotool.com/download/ster ... _BETA2.exe
VST version: http://www.stereotool.com/download/vst_ ... _BETA2.dll


If there are still problems in the audio of BETA2 that aren't present in BETA, then there's a problem in the BETA2 processing.

If the problems are present in both versions, I just need to fix the Trumpet filter for BETA2.

As mentioned earlier, you might need to turn AGC/Pre Limiter off to be able to compare the two.
And don't forget to lower the 10th 'soft limit' slider in Multiband in BETA2!

And there's more! I just checked, and in v4.22, the highs above the lowpass frequency are filtered out at the lowpass frequency + 400 Hz before Multiband. In the latest version, they are filtered out at exactly the lowpass frequency.

The consequence of this is that there are LESS highs in the final band - so they come out louder.

In v4.22, there was a 2nd lowpass filter somewhere after Multiband that filtered out the highs between 'lowpass' and 'lowpass+400Hz'. (That's also happening in the latest version, but it's having far less effect.)

This might explain the +0.5 dB in the high frequencies that I saw. (This is a separate issue from the 'highest band soft limit must be set lower' issue, which results in even more highs in the end result).

I'll fix this tonight.

Please hold the testing (if you are testing). There are more bugs in the new multiband implementation. Will try to fix them tonight also.

now after reading all this and additional tests .. now i think main problems are in multiband. But sometimes differences are so small on many songs.
10th soft limit slider .. i allready put it lower days ago .. if i play same song thru 4.22 and beta2 (not trumpet - same preset - fm output) i see in 10th meter is bit higher, and sometime 9th.
Also in v4,22 Final Limiter much more goes in "red", but i don't hear difference there.

FM Output:
i said this erlier.. to have same maximized sound on FM as in 4.22 (with fm overdrive 101%) .. in beta2/3/3a .. i must set FM Overdrive to 102-102.5%.

edit + lol

New version with fixed Multiband is building now. Output volume for all frequencies is now within 0.1 dB of that of v4.22 (except for the improved clipping).

Stereo Boost + Noise Gate + Lowpass + Highpass + Multiband incl clipping + Final Limiter (at 1.50, deep in the red) stays within 0.1 dB of the v4.22 output for higher frequencies (above 400 Hz or so), at lower frequencies there's a bit more difference, probably due to the highpass filter. (above 48 Hz the new version is slightly louder, below 48 kHz v4.22 is louder, with highpass filter frequency set to 30 Hz).

Adding Loudness makes the difference a bit bigger, probably mainly because the output level is slightly lower (-0.1 dB) in the new version compared to v4.22. This is most likely caused by a filtering step that I had to calibrate again for the new version.

I should add to that that both Final Limiter and Loudness have a more constant output level in the new version. That _could_ indicate that it causes more distortion (but based on my measurements it should actually have less distortion).


Short explanation of what was wrong:
■ In v4.22, when both Multiband and Lowpass were active, I filtered out frequencies above 400 Hz ABOVE the Lowpass frequency before Multiband. Only at final steps I filtered at the exact lowpass frequency. In BETAx, I filtered at exactly the lowpass frequency already before Multiband. Result: A 400 Hz band that was gone, resulting in a lower measured volume in the highest band. The remaining frequencies then came out louder.
■ I accidentally turned off a filter that I wrote a lot of versions ago to prevent annoying sounds caused by very big output level differences between bands (basically a part of the 'neighbour' bands is added to the signal). This didn't work anymore in the new version... I never noticed it because the difference in absolute output is very small - but, depending on the type of sound, it could get big for the highest band. Filter is now working again, but I fear that the CPU load went up quite a lot due to this, so I'll have to optimize it later. (For now, I just want the quality to be good).

Uploaded again as BETA2.


After doing some quick measurements it seems that in the new implementation, AGC (dualband) 'soft limit' must be set about 13% higher than in v4.22 to reach roughly the same Pre Limiter/AGC output levels. I don't know if the highs and lows are similar as well, but the RMS volume of a sample set of 15 minutes of music ended up at the same level (difference 0.02 dB).



Edit: Just measured the frequency response (took some time to analyze the signal...). Based on the default settings of AGC compared to default settings of Pre Limiter, the bass sounds (below 250 Hz) are 0.8 dB louder, higher frequencies are 0.5 dB softer. Setting the relative volume level to 85% (instead of 70%) makes the frequency response similar to that of the old pre limiter, again at least for the example sounds that I put in. In that case the 'soft limit' slider must be set a bit lower (I would guess around 0-5% above the level for v4.22, but I'll have to measure it to be sure).

I don't know if it's really useful to further pursue this. My impression so far was that setting the highs at 70% sounds better than setting them higher, because I feel that for some recent songs the bass is lacking in that case.


Edit: Ok, I've tried +0% with 85%, and now the bass is almost at the same level (-0.2 dB), but the highs are at -1.5 dB.

- For now sound is very nice and clear. Highs are softer, maybe more then should but this is somehow fine. i'll must listen more on FM...
- In AGC i keep first slider to ~58.00 while in multiband Compress/Limit is at "half".

- btw.. maybe dumb question, but what is the two sliders in Multiband (under the Output Limit slider)?

- Hans can you try this song http://www13.zippyshare.com/v/60281486/file.html

Will do that tonight. What should I hear?


About the slightly softer highs: I suppose I could change that (or even make it configurable...)

Extremely low bass with high level from song, it confuses stereo tool, especially if loudness is lower or off.

edit: Second band in AGC is lifesaver for this song

I hear what you mean... Version 4.22 does nearly the same thing. Dualband AGC does indeed more or less solve it (now you know why it's needed ).

If I turn AGC off completely the problem is gone.
If I enable dualband AGC, it's also nearly gone (the sound is less 'warm' when the loud very bass is there, but no huge drop in volume here).

Possible solution: Ignore VERY low frequencies when calculating the RMS volume.