Stereo Tool

After getting some complaints I noticed that the CPU load had increased dramatically since version 3.21.

Version 3.27 is still 6% slower than version 3.21 (the sound quality is a lot better), but it's 48% faster than version 3.26. (In other words: The CPU load will be about 30% lower than in version 3.26).
Also, an extra switch has been added to the Louder filter. If you turn "++" on, you get the same effect as in older versions. If you turn it off, the sound will be slightly softer, but the dynamics are preserved much better and the sound stays much closer to the sound without Louder filter.

all run great , thank you hans

I have switched back to 3.22 as I find it the better when it come sto cpu utilization
in fact it uses 25-30% less then 3.24, 3.25, 3.26 or 3.27.

With 3.22, I am using on avg about 27.83 - 29.4 % where in particularly with 3.27 49.81 - 51.35% which is a significant difference.

I either case I am using the same settings.

I am using Stereotool with SAM and have noticed that whenever I start SAM a conenction is made back to sterotool.com and just sits there until SAM4 closes in a close wait state.

Why is there a process back to seterotool.com and why does after the process is opened remain in the close wait sate? Does this happen with winamp also? What is the of the process back to stereotool.com?

It remains in this state until SAM4 is shut down.

Can you send me the settings that you are using? I've done a lot of measurements which show Stereo Tool 3.27 to perform nearly equally to 3.22...

About the "call home": Stereo Tool checks for new versions. It can report that a new version is available in the "VISIT WEB SITE" button - that's why it's so big. I'm planning on using it when a big improvement is ready (and I've been a bit hesitating to do so lately because of the high CPU usage). I don't know why the connection stays open - I'll check the code (might be missing a close somewhere).

Here's what I'm measuring:

Preset: FM Loud Europe

Quality HIGH: 3.27 is 40% faster than 3.22
Quality MEDIUM: 3.27 is 6% slower than 3.22
Quality LOW: 3.27 is 5% faster than 3.22

Without Loudness:

LOW: 3.27 is 2% slower
MEDIUM: 3.27 is 1% slower

Versions 3.24-3.26 were indeed 30% slower (which would give numbers like the ones you are reporting), due to a bug in the final limiter and hard limit filters which made things really slow. But I'm very surprised that 3.27 is also slow on your pc.

Of course, different CPU's (with different cache sizes etc.) might behave differently. (What CPU do you have?). Also, how did you measure the CPU usage? The numbers reported by Windows (Task Manager etc.) are not always reliable.

Another measurement however did show that 3.27 is 16 (!) % slower than 3.12 using the same settings, Loudness at 1.50, quality Normal. (The sound quality in 3.27 is a lot better than in 3.12 though). Without Loudness the difference is smaller (7%). I will run a test on my old PC (which isn't mine anymore, but I can run a test on it tomorrow) to see what it does on older/slower hardware with less cache. The increase does make some sense (I've added a lot of extra filters to increase the sound quality, for example the lowpass filter is now used 3 times instead of once to increase the brightness of the audio, the final limiter is rewritten for better sound with less artifacts, Hard Limit is split into two passes instead of one, and extra post-filtering is added to the Loudness filter to remove artifacts that it caused. Still, it might be time to look into the performance of things once again. I'll get back on this after my measurements tomorrow...

I have just done a measurement on my old pc, and the difference is not bigger than expected. Versions 3.1x used about 80-84%, 3.27 uses around 90-93%. So that's about 12% more.

Note that this is especially the case when "Direct Sound Card Output" is enabled. Otherwise it's about 40%.

This is on a 5 year old 2.4 GHz Pentium 4 pc.

The difference in CPU load is roughly identical to that of my own pc. So it looks like the load is consistently about 12-16 % higher for version 3.27 than version 3.1x.

Still, I'll take another look at the performance after I finished the VST plugin version. But I don't think I can improve it a lot more (I could however turn off some of the extra filtering in Low quality mode).

I am using " kernal streaming output " as this lets me talk directly with my sound card.

Have you tested it with " kernal streaming output " and then compared between them.

My sound card is an x-fi elite pro which allows me to use kernal streaming for a better sound quality.


I am getting using 3.22 between 20-25% lower than using 3.27 hence my switching back to it. I am averaging 25-32% whereupon using 3.27 I get consistently 45-51% which is significant.

I am not using task manager as I know it yields errouneous resluts.

Hi Mapleleaf,

Which parameters or parameter set are you using? That might make a considerable difference...
And what type of CPU do you have? (Most importantly: Does it support SSE2?)

I have an Intel Pentium 4 3.2Ghz and ye sit does support SSE2.

Have you done a comparison between Direct and Kernal ?

Since my sound card allows me to use ' Kernal streaming' for my output in SAM4, I do get a far better quality in audio.

I will email you a copy of my *.sts that I am using and also screenshots BUT this will take a bit of time as I am just finishing a project to meet the deadline.

Hi Mapleleaf,

Kernel streaming is something outside of Stereo Tool control, so it cannot explain the difference between the two versions. Direct Soundcard Output in Stereo Tool is a mode used to feed FM transmitters, which sends out a 4 times higher sample rate. Part of the processing is also done on that higher sample rate, which causes a higher CPU load (although I wouldn't expect it to be as high as I measured).

I'm currently busy looking through the processing steps that I'm using, I've already found one extra filter that had no effect on the sound (I've removed it). I'll post an update later tonight which should have a lower CPU usage than 3.27.