Stereo Tool

This is what I've been doing for a while in hardware. I find it works decently well. There are still some transients you have to be weary of, however. Additionally, a fast drop across a multiband processor is less noticeable than a fast drop with a single/small multiband compressor. Transients tend to be limited to a limited frequency range, relatively speaking. As long as the rest of the processor catches up quickly enough, this can be almost transparent.

Time to play!

Here's an interim build of my HE-AAC settings using BETA3:

2ND AGC is enabled. Config:
3 Band
Target Output: 12.50 (-18.35 dB RMS)
Remove remaining peaks above: 150%
Gating level: OFF
Channel separation: 75% (12dB)
Up Speed: .00000/.00000 (Default)
Down Speed: .0003/.0003 (Default)

So far, so good. Very smooth. Here's the rest (and attached, for convenience):

Treble is reduced for HE-AAC. This all can be adjusted easily, and the dynamics will follow accordingly. For the most part, any radio station I've ever configured sounds pretty close to this. I like perceived dynamic range.

[Common]
Pre amplifier=5
Post amplifier=0.899999976
Extra loudness=1
Extra loudness - Very deep bass relative clipping level=0
Extra loudness - Allow extra lows (advanced)=0
Extra loudness - Allow extra highs (advanced)=0
Extra loudness - Punchy=0.284999967
Extra loudness - Max Punch peak=0.75
Extra loudness - Deep bass protection - Enabled=0
Extra loudness - Deep bass protection - Difference=0
Extra loudness - Deep bass protection - Value=0.638750076
Extra loudness - Forcibly remove DC component caused by Loudness=0
Bass protection - Enabled=1
Bass protection - Two steps=0
Bass protection - Clip level=140
Bass protection - Deep bass boost=0
Bass protection - Raise clip level if highs get above=150
Bass protection - Maximum clip level raise=35
Hard limit output=1
[Noise Gate]
Enabled=1
Difference=0
Noise level=3.849999905
Relative noise gate level - Band 1=1
Relative noise gate level - Band 2=1
Relative noise gate level - Band 3=1
Relative noise gate level - Band 4=1.150537491
Relative noise gate level - Band 5=1.352941036
Relative noise gate level - Band 6=1.409638405
Relative noise gate level - Band 7=1.150537491
[Singleband Compressor]
Enabled=1
Difference=0
Maximum volume=4
Attack speed=0.000001098
Decay speed=0.000763275
Above Top Limiter=0.199999988
[Pre Compressor]
Enabled=1
Difference=220
Delay enabled=0
Maximum volume - Band 1=22500
Attack speed - Band 1=0.000001125
Attack speed - Band 3=0.000001284
Decay speed - Band 1=0.0010933
Decay speed - Band 3=0.0010933
Dual band=1
Third band=1
Relative expected volume of highs=0.5
Relative maximum volume of highs=1.199999928
Relative maximum volume of very highs=0.149999991
Remove remaining peaks above (relative to maximum volume)=5
Maximum expected dynamics strength for smoothing=2
Enable smoothing=1
Smoothing time=12.529999733
Gating level=123201
Use ITU-BS.1770=0
[Multiband Compressor]
Enabled=1
Difference=0
Delay enabled=0
Maximum volume - Band -1=5950
Maximum volume - Band 0=7250
Maximum volume - Band 1=7799.999511719
Maximum volume - Band 2=7500
Maximum volume - Band 3=6350
Maximum volume - Band 4=4200
Maximum volume - Band 5=3750
Maximum volume - Band 6=4650
Maximum volume - Band 7=3600
Maximum volume - Band 8=3350
Attack speeds linked=1
Attack speed - Band -1=0.000005229
Attack speed - Band 0=0.000005229
Attack speed - Band 1=0.000005229
Attack speed - Band 2=0.000005229
Attack speed - Band 3=0.000005229
Attack speed - Band 4=0.000005229
Attack speed - Band 5=0.000005229
Attack speed - Band 6=0.000005229
Attack speed - Band 7=0.000005229
Attack speed - Band 8=0.000005229
Decay speeds linked=1
Decay speed - Band -1=0.000089597
Decay speed - Band 0=0.000089597
Decay speed - Band 1=0.000089597
Decay speed - Band 2=0.000089597
Decay speed - Band 3=0.000089597
Decay speed - Band 4=0.000089597
Decay speed - Band 5=0.000089597
Decay speed - Band 6=0.000089597
Decay speed - Band 7=0.000089597
Decay speed - Band 8=0.000089597
Above Top Limiter=0.75
Clipping enabled=1
Postprocessing enabled=1
Relative clip position - Band -1=-1
Relative clip position - Band 0=-1
Relative clip position - Band 1=-1
Relative clip position - Band 2=-1
Relative clip position - Band 3=-1
Relative clip position - Band 4=-1
Relative clip position - Band 5=0.754385829
Relative clip position - Band 6=0.801801801
Relative clip position - Band 7=0.65289247
Relative clip position - Band 8=0.90476203
Final limiter value=1
Final limiter decay speed=0.999000013
Final limiter clipping=20
Equalizer enabled=0
Equalize before multiband-compression=0
Equalizer position - Band -1=1
Equalizer position - Band 0=1
Equalizer position - Band 1=1
Equalizer position - Band 2=1
Equalizer position - Band 3=1
Equalizer position - Band 4=1
Equalizer position - Band 5=1
Equalizer position - Band 6=1
Equalizer position - Band 7=1
Equalizer position - Band 8=1
[Stereo]
Enabled=3
Stereo AZIMUTH Enabled=1
Stereo Image Enabled=0
Stereo Boost Enabled=1
Stereo Boost Excessive Reverb Protection Enabled=0
Delay enabled=0
Difference=0
Center bass=0
AZIMUTH limit=25
AZIMUTH change speed=0.199999988
Image phase amplifier=1
Image phase amplifier maximum angle=180
Image phase amplifier maximum separation strength=100
Image width amplifier=1
Extra phase shift=0
Mono or stereo only=0
Stereo Boost strength=10
Stereo Boost maximum amplification=0.799999952
Stereo Boost multiplier=0.899999976
Stereo Boost - Never reduce below original level=0
[Channel Delay]
Enabled=0
Left Delay=0
[Output Filter]
Enabled=0
Lowpass filter=15000
Highpass filter=25
Highpass filter linearity=Auto
[Bass Boost]
Enabled=0
Strength=1
Start lowpass frequency of input filter=0
End lowpass frequency of input filter=250
Start lowpass frequency of output filter=250
End lowpass frequency of output filter=350
Clip level=1.5
Clip level is relative to estimated maximum level=1
Preserve louder peaks=1
Difference=0
[Phase Rotator]
Enabled=0
Start frequency=3500
Start speed=1350
Number of rotations=6
Smoothness=0.985999942
[Final Limiter]
Enabled=1
Difference=0
Pre-amp=1.5
Pre-limiter enabled=1
Pre-limiter relative volume=1
Reduced CPU processing=0
[FM Transmitter]
Enabled=0
Pre-emphasize=0
Pre-emphasis time=50
Output is pre-emphasized=0
Stereo encoder enabled=0
RDS encoder enabled=0
Pilot signal volume=9
RDS signal volume=4.5
FM overdrive volume=1
RDS PS text=2s:STEREO/2s:TOOL/<1=1.5s,2..-2=2t,-1=1.5s:WWW.STEREOTOOL.COM/|\W"PLAYING %artist% - %title%        "/1s:\T2h\:\T02m\:\T02s/1s:\T2h\:\T02m\.\T02s/1s:\T2h\:\T02m\:\T02s
RDS RadioText text=60s:Stereo Tool: Professional Audio Processing - www.stereotool.com/30s:Stereo Tool by Hans van Zutphen, 1999-2008 - www.stereotool.com/30s:\W"%artist% - %title% - %album%"
RDS PTY=0
RDS PI=65535
RDS Alternative frequency 1=0
RDS Alternative frequency 2=0
RDS Alternative frequency 3=0
RDS Alternative frequency 4=0
RDS Alternative frequency 5=0
RDS Alternative frequency 6=0
RDS Alternative frequency 7=0
RDS Alternative frequency 8=0
RDS Alternative frequency 9=0
RDS Alternative frequency 10=0
RDS Alternative frequency 11=0
RDS Alternative frequency 12=0
RDS Alternative frequency 13=0
RDS Alternative frequency 14=0
RDS Alternative frequency 15=0
RDS Alternative frequency 16=0
RDS Alternative frequency 17=0
RDS Alternative frequency 18=0
RDS Alternative frequency 19=0
RDS Alternative frequency 20=0
RDS Alternative frequency 21=0
RDS Alternative frequency 22=0
RDS Alternative frequency 23=0
RDS Alternative frequency 24=0
RDS Alternative frequency 25=0
RDS TP=0
RDS TA=1
RDS Music=1
RDS Artificial Head=0
RDS Compressed=1
RDS Dynamic PTY=0
RDS RadioText Enabled=1
RDS ClockTime Enabled=1
FM Transmitter Calibration Enabled=0
MPX Oversampling mode=Follow main Quality setting
FM Transmitter Calibration - Volume 20 Hz=1
FM Transmitter Calibration - Volume 40 Hz=1
FM Transmitter Calibration - Volume 80 Hz=1
FM Transmitter Calibration - Volume 160 Hz=1
FM Transmitter Calibration - Volume 320 Hz=1
FM Transmitter Calibration - Volume 640 Hz=1
FM Transmitter Calibration - Volume 1280 Hz=1
FM Transmitter Calibration - Volume 2560 Hz=1
FM Transmitter Calibration - Volume 5120 Hz=1
FM Transmitter Calibration - Volume 10240 Hz=1
FM Transmitter Calibration - Volume 19000 Hz=1
FM Transmitter Calibration - Volume 23000 Hz=1
FM Transmitter Calibration - Volume 28000 Hz=1
FM Transmitter Calibration - Volume 33000 Hz=1
FM Transmitter Calibration - Volume 38000 Hz=1
FM Transmitter Calibration - Volume 43000 Hz=1
FM Transmitter Calibration - Volume 48000 Hz=1
FM Transmitter Calibration - Volume 53000 Hz=1
FM Transmitter Calibration - Volume 57000 Hz=1
FM Transmitter Calibration - Phase shift 20 Hz=0
FM Transmitter Calibration - Phase shift 40 Hz=0
FM Transmitter Calibration - Phase shift 80 Hz=0
FM Transmitter Calibration - Phase shift 160 Hz=0
FM Transmitter Calibration - Phase shift 320 Hz=0
FM Transmitter Calibration - Phase shift 640 Hz=0
FM Transmitter Calibration - Phase shift 1280 Hz=0
FM Transmitter Calibration - Phase shift 2560 Hz=0
FM Transmitter Calibration - Phase shift 5120 Hz=0
FM Transmitter Calibration - Phase shift 10240 Hz=0
FM Transmitter Calibration - Phase shift 19000 Hz=0
FM Transmitter Calibration - Phase shift 23000 Hz=0
FM Transmitter Calibration - Phase shift 28000 Hz=0
FM Transmitter Calibration - Phase shift 33000 Hz=0
FM Transmitter Calibration - Phase shift 38000 Hz=0
FM Transmitter Calibration - Phase shift 43000 Hz=0
FM Transmitter Calibration - Phase shift 48000 Hz=0
FM Transmitter Calibration - Phase shift 53000 Hz=0
FM Transmitter Calibration - Phase shift 57000 Hz=0
[Direct soundcard access]
Enabled=0
Device ID=Microsoft Sound Mapper
Volume=1
Buffer size=1
Send to Winamp=Nothing
ASIO Override channel 1=4
ASIO Override channel 2=5
[Low latency output]
Enabled=0
Device ID=ERROR: INVALID DEVICEID
Volume=1
Buffer size=0.079999998
ASIO Override channel 1=2
ASIO Override channel 2=3

ok.. this L/R waveing is solved...and i think for all tracks this mono control is better.

Hans, is it possible to have values in ms for down and up speed...
i know that it's "Pain in the a**" .. i think that this with two AGC's can be good .. but it so hard to tune it .. one "pixel" up or down of 1st or 2nd agc changes a lot. With exact values in "ms" things can be easyer.
For multiband too.
?

I tried to fill in values in ms / 6 dB, but I discovered that the AGC filter is behaving very differently from what I expected.. That might be bad (I'm not sure yet).

At lower volume levels (eg. more compression is taking place), the up speed is relatively higher. I think it should (and expected that it would) be identical.

Did i undarstand good...
Up speed i changed if volume is changed.?

Thats really not good!

This is VERY interesting!

I've tried to make the volume effect more constant, but that made the voice problem a lot worse.

Bojcha has sent me a PM a few days ago about reducing the up speed when the current level and target level are close to each other.

Until now, the volume increment was very fast if the volume was very low, changing with an asymptote to 100%.

I've now changed this; instead of an asymptote that ends at 100%, the asymptote now ends at the new target volume level at the current position (so if the level doesn't change much, the volume stays almost the same).

Result:
- Volume of already compressed tracks is far more stable than before.
- Between voices, if there's not really a lot volume difference, the volume doesn't climb much, so it doesn't drop much when a new voice starts either.
- Relatively silent parts of tracks are not boosted as much as before.
- Big volume differences are handled similarly to before, which is good.
- And for the same track, the up speed is now identical regardless of the input volume (unless it's really low of course).

I'll start a build now and upload the new version tomorrow morning.

New version:
Winamp DSP plugin: http://www.stereotool.com/download/dsp_ ... BETA3A.exe
Stand alone version: http://www.stereotool.com/download/ster ... BETA3A.exe
VST version: http://www.stereotool.com/download/vst_ ... BETA3A.dll
VST version (No SSE2): http://www.stereotool.com/download/vst_ ... BETA3A.dll
Command line version: http://www.stereotool.com/download/ster ... BETA3A.exe

Notes:
- OOPS! I see that I didn't remove the code that I added to display ms / 6 dB values instead of .000001 - but the values don't make any sense .
- For the same 'up speed' behavior, you might need to set 'up speed' slightly higher than before. (But I don't know if that's really necessary)
- Difference in dB ITU-1770 is a big higher now (unless you set 'up speed' slightly higher), but not that much (for the worst case situation, "Gimme Gimme Gimme" vs. very strongly compressed song, 1 channel, at default settings the difference increased from 2.1 to 2.8 dB; if you use 2 channels the difference is about 0.7 dB).

Cause of not working Loudness in last few versions (BETA2/3/3A) found - and solved. I forgot to release memory at 3 different places (!), Stereo Tool's memory manager then tries to log that to a file - and if it has no access to write to that file, it just stops. Anyway, this situation should NEVER occur - and it's solved now. Update coming later.

If you turn on 'two steps" there is no bug with loudness

So, is it possible to convert values in agc and multiband to ms (milliseconds) instead of *.00001* ?

Probably not, the volume response is still not constant and now also depends on both the current AGC dynamic volume level AND the current input level....

As I said before, I tried what happens when I _do_ make the response constant, and it really sounds a lot worse - far more volume effects.

What I could do is calculate the 'half-time' based on a constant difference between current and input level. Say, if the input volume drops to 25%, how long does it take before the AGC volume level has doubled. But I'm afraid that that would only make things more confusing... (Better would be: The speed increment if the input volume drops to 50%, and then the time it would take to reach 50% of the volume, based on the volume increment at the moment where the volume drops... Will have to think about this )


By the way, have you tried the new sound?