Theo Verelst Local Diary Page 75

I've ditched the usual header for the moment, I think it doesn't help much anyhow.
This page is copyrighted by me, and may be read and transferred by any means only as a whole and including the references to me. I guess that's normal, the writer can chose that of course, maybe Ill make some creative commons stuff one day, of course I have made Free and Open Source software and even hardware designs available!

Home (alt)   Tripod Home   Local Diary list   Tripod Diary List   Previous Diary

This page is under construction, so check back later, too.

May 9, 2009

HD screendumps and pictures again... (or: still)

Well I think a modern browser can instead of blow up text and images also shrink them, to fit on your screen. Just like one cannot make more resolution in general except with progressively coded video, one often cannot easily make chocolat from shrinking a lot of web things and other pixel-messed pictures down. When one works decent with the visual materials, especially shrinking should not be much of a problem, like I often check how my 1080 (i) movies look in 720 size, for web use:

That, of course can be an excuse for poor resolution big formats, which aren't as accurate as their resolution would need, but that's not really the case, the camera I used mostly for years is an actual 1080(/50i) Sony HD cam, with 2.2 megapixels, though probably the scaling down of the PureVideo player hardware mp2 player is pretty likable to get more pixeljuice from the 'footage', I can't examplify that, because the overlays can't be screen dumped, and mostly it works for moving video.

Impulse Reverb

Impulse reverb means a reverberation effect applied to an audio source which

Audacity showing the self compiled (not written!) Freeverb 3 on Linux (Fedora 10/64) which is a different thing then freeverb version 3 from the ladspa plugins, it´s a heavy number crunching fftw3-library based reverb and compressor, and contains a impulse sample based convolution reverb, which moreover also works in real real-time.

I´m not a total impulse reverb fan at all, in fact I´m a bit known to be aware of the theoretical error to actually believe in it (the system isn´t linear and time-invarying as is required for sampling theory and system theory which uses convolution integrals, so the theory and it´s nice proofs really doesn't hold).

On the other hand, for non-challenging a bit characteristic reverb impressions, or as in freeverb3 in combination with more data-storing and processing other reverbs, it is fun to play with to some extend.

3 short 24 bit stereo example wav files (I used very short pieces or Neil Diamond, Golden Earring and my tryout piece from lately: Tipatina), 256kb/s mp3 (which sounds less lively in spite of the good bandwidth):

  audaci1 [.wav] (6.3 MB)    audaci1 [.mp3] (0.8 MB)
  audaci2 [.wav] (6.3 MB)    audaci2 [.mp3] (0.8 MB)
  audaci3 [.wav] (8.8 MB)    audaci3 [.mp3] (1.1 MB)

Patching up the mid range in compressed CD tracks

Making use of the averaging and other properties of the FFT based Jamin 1024 band graphic equalizer and some 10 band equalizer LDSPA (Linux) plugins for filtering the low and the high end out and into a seperate, slightly delayed signal path, I tried to repair the sligtly undersampled (not so many bits) and brittle compressed mid range of a number of audio CD songs:

like above, in principle the signal path can be seen from those windows and the jack connection window in the upper right hand.

Because the experiment was kind of succesfull, though I suppose nothing in this ballgame is perfect, I also tried adding reverb of both the impulse (there are quite a few impulse responses in stereo wav form on the internet nowadays) and the algorithmic kind combined, to keep the sound from getting dull very soon.

Overdoing the reverb, especially the hall (various size) like sounding resonating algorithmic ones, and even worse when adding a little compression after that brings back a lot of the original (before horrifying re-mastering techniques) sound richness, and a live feel when adding yet another (I tried plate simulation) type of reverb after the after-reverb compressions.

Can make everything from Elvis and Domino to Kiss and Black Crowes to live, and when equalized right (easily up to 20 dB (!)) for the lowest notes (and remember, I use  sub woofer which starts to work under 40 Herz, and has more than a few hundred RMS Watts to do so, so I´m not compensating for the sound reproduction system but only for the flat-compressed, low-bereft CDs), the result contains astonishing low musical notes up to disco boom without the addid sh*t sounds, like a rock concent, without requiring enourmous power. Excercise to the reader: explain this...

Of course for the attentive and knowledgable audience, it is reasonable to wonder if after the mid upgrade to 24 bits with FFT based analysis (I know this is more involved than can be summed up in a few sentences), a decompression step (like I described in previous pages) would be appropriate. The answer is: of course, though getting all the dynamics right is possibly a bit utopical. In practice the results are great, but I suppose it would be better to get some good SACDs or so.

I´ve made one example , which is Jezebel, from Sade.

Physical Modeling with the Stanford STK

Great! I´ve compiled and used the STK kit on two machines, and played with the sounds, I found out  the demo required a program change to make sound, but it works quite well, and it is nice to play physically modeled sound, even if the complexity is modest. Such a musical relief after so many samples.

A short example of the clarinet sound with a impulse reverb on it:

     stkpm1 [.wav] (4.3 MB) stkpm1 [.mp3] (0.8 MB)

I played a lot more on another machine and a good keyboard, making more subtle and actually some quite funky sounds with the string instrument simulated sound, not as great and deep as my own, but very musical. I didn´t record this. (Yet.)

Could I predict certain DSP types would perish because of their musical boringness ?

Sure. Easy.

How: forget 3/4 of the usefull aduio spectrum, compress sound represenetations to far below the clear inaudible encoding limit, keep defending transforms and simplifications which are certain not to add all too much musical value, and are always not good or even stupid, keep peoples interest in seless standing waves and undesirable ways of (ab-)using them, presume all kinds of "new" "musical" and "production" algoithms are new and therefore not hampered by th aforementioned to be actually total crap.

Oh, and some antichristal "persons" who happen to have to take the place of me and the real musicans doing much more interesting, musical and complicated processing already in the 70s by the help of very dark groups of "people", gathered up false trust and probably loads of (corruptly used) money from ditto are not going to get my vote for being very probably the new replacement of the the Rolling Stones and Stock Aitkin and Waterman. Seriously.

Keep believing them, of course....

One day the new music will become great. Sure M*f*.

I put this question up on the Audio Precision website:


      Would you also know about equipment (possibly software) which deals with things like

    signal shift-induced distortion specs, or FFT or other frame based distortion specifications?


Hydrogen drumcomputer

I´ve also compiled (though it is also available using yum, which works) Hydrogen, a GPL Open Soure program making a software drumcomputer work with a number of drum sets, in my case also on the FC10/64 Linux system, under Jack.

Here are example with a microphone speaking over the impule reverb, and starting a piece of the drum demos, also sounding over the reverb (which is a hard test):

  hydrog1 [.wav] (3.8 MB)   hydrog1 [.mp3] (0.5 MB)
  hydrog2 [.wav] (3.8 MB)   hydrog2 [.mp3] (0.5 MB)

The Tcl to automatically make the links shown for this page from the audio files:
foreach i [lsort -dict [glob Ldi75/*.wav]] {
foreach k [lsort -dict -decr [glob [file rootname $i].* ]] {
set j [file tail $k]
puts -nonewline "<a href=\"$k\">[file rootname $j]</a> \[<i>[file extension $j]</i>\]
([format "%0.1f" [expr [file size $k]/(1024.0*1024)]] MB) "
puts "<br>"

Some more TV I grew up with

I'm from Holland, so let me say some nice things about people in the dutch past, too, so the contrast is more clear and credible.

FIrst, some of the children TV shows, look at the fragments, there somewhat representative, even if you don't understand dutch, they should be clear enough, and these were major shows, not marginal and unimportant ones.

This particular showI didn't really see much:

this one not either:

Very common, daily show:

I guess I saw this show regularly:


And some great music:

Of course his hit Saturday Night was more international hit, even in the billboard top ten I think.

At highschool I regularly saw the schoolparty bands (in fact I suppose all of them, and that was in the Hague) and I recently found out the band Vitesse was with him in it, I didn´t know that, it was pretty good music though, and indeed (rock and pop) music, not sh*t.

 Het land van Maas en Waal

 Ben ik te min

 Meneer de President

Heavy Rock

Live and let die:

Baron Samedy versus Mardi Grass

End 80's music

and I mean music, not sh*t.