Powerballalyzer now available!

[Pako]

Hi,

cool program! It didn't miss any of my scores that I tested.

May I suggest following features for next release:

- context menu, so that you can analyze sound files by just right-clicking the file.
- support for different sound settings (mono/stereo, x bits, x Hz); or internal conversion before the analysis starts.
- faster analysis. I found it rather slow (on my 2GHz laptop). Couldn’t the calculations be faster? Maybe use of MMX/SSE/SSE2/SSE3 support?
- some kind of curve recognition? Maybe detecting possible curves (each one indicating its maximum score) listed in order (most probable first), so that user can choose?

I don’t have much idea about sound, graphics and user interface. I’m also a programmer, but I’m expert in low-level stuff. Could you distribute the source code? I could do reverse-engineering of the code, but don’t have time. TheForce: would that be illegal?

Regards,
Pako.
_________________
REGULAR:

speed = 15136 ; 13606(L)
30s = 6662
90s = 18244 ; 17593(R) ; 15558(L)
dual = 27113
reverse = 14233(R) ; 12900(L)

SOUNDBALL:

speed = 14636

METAL:

speed = 13406
90s = 15465

[Publicités]

[mroek]

I'm not a big fan of context menus, I find that too many programs have a tendency to add themselves to the context menu. But I suppose it could be made a user choice. Supporting different sound file formats would certainly be possible, but for now I haven't really cared, as it is pretty easy to convert the files in audio software before analysis.

I agree it is slow, but I haven't done any great efforts to speed it up. The slowness stems mostly from heavy use of the sin() and the cos() functions. Speedier sine and cosine could probably be used, but possibly at the expense of accuracy.

Curve recognition is difficult, although probably possible. The human brain is much better at this, and can in an instant identify the top of the curve. Humans can also easily interpolate (visually) if the curve has holes or other anomalities.

I don't want to release the source, partly because it is quite messy, and partly because I just don't feel like it..

[mroek]

I've now increased the speed significantly. The current beta (not available yet) does the audio analysis more than 4 times faster than the previous version. It could probably be made even faster, but that's a bit more work, and not worth it in my opinion.

Also note that the longer the audio file, the longer the analysis takes.

[donkapone]

wow... with that speed (and if you could make it even faster) it would be easier to make it realtime...
_________________
sin cera, Arturas

[mroek]

Yes, I believe that real-time is within reach, but I am unable to see the benefit.

Getting an *accurate* RPM fully automatic in real-time is probably extremely difficult. Getting a *rough* estimate of RPM in real time is probably doable, but is it of any use? If it's not accurate, what's the point?

On a side note, I've also thought about adding support for different WAV-formats, but I'm not sure that's much use either. I mean, to verify scores from videos you'd have to use some other piece of software to get the WAV in the first place, and doing the output conversion in that software is not much work. If you are just recording sound, then you just specify the correct format to begin with.

[Obiw4n]

It could be great if the length of the possible analysis was increased to 90s minimum.
To analyse 90s runs of course. Here I can only see 30s max.
_________________
Help Gyrotek when you buy your Powerballs, use our banner: International, France, Thanks!!

[mroek]

Ok, here's a new beta version with a few improvements. Most notably, the analysis window has changed from grayscale to color and the program is faster. The program still only accepts one type of WAV, but the length of the file can now be up to 100 seconds. The program will now fit the entire sound to fill the output window. You can also add the program to the right-click context menu by adding a shortcut to it in the "SendTo"-folder. See the readme file for more details.

Download link:
Powerballalyzer ver 0.2

I am thinking about adding some kind of real-time capability, but it might not be very useful. We'll see.

As always, comments are welcome and appreciated. With no comments it feels as if the work is a total waste of time...

[Obiw4n]

Thank you a lot for this second version!

good things:

The coulours enable a better viewing, better contrast.
The 100s max lenght is greatly appreciated.

Now i'd like to be able to make a "zoom" on a curve to see precisely a peak for example. could be useful I think. and the frequency will have to appear in the zoom window too of course.

Don't worry, since you released your prog, I use it almost every day to check every video record!!!
_________________
Help Gyrotek when you buy your Powerballs, use our banner: International, France, Thanks!!

[mroek]

I have been thinking about adding some kind of zoom-functionality, it might come in future versions. The project has become a bit more work than I thought, and much of the old code had to be rewritten. I'm still not sure there will be enough users to warrant spending a lot of time on it.

But as I said, time will show... :-)

[Obiw4n]

Take your time, the present program is sufficient enough for what we basically need.
_________________
Help Gyrotek when you buy your Powerballs, use our banner: International, France, Thanks!!

[donkapone]

Don`t worry about the usability of program - i use it a lot too... just now it doesn`t work with my sounds... it fills with the blue screen and that's all... i liked the grayscale version more...
_________________
sin cera, Arturas

[Edouard]

Very good program

Thanks a lot
_________________
Speed:
250Hz: 14702 (R) 15079 (L)
350Hz: 12710 (R) 11921 (L)
Dual: 27628

Endurance:
250Hz 30s: 6754 (L+R)
250Hz 90s: 18803 (L+R ed tech)
350Hz 90s: 15824 (L+R ed tech)

[Adrena1in]

[mroek]

[donkapone]

ok, now it works... but the grayscale version was more accurate i believe... no?
_________________
sin cera, Arturas