Audio Experiments

Thu, 1 Aug 2013

Time for a quick recapitulation:

I've made a sound recording of myself saying "hello" and stretched out the waveform in Audacity so that all the peaks and troughs are clearly visible.

I traced the waveform in Illustrator then wrapped it round in a circle.

I constructed a paper horn in the hope that I would be able to play the sound back.

The end of the horn has a read head which follows along the waveform hopefully playing back the sound "hello" from the paper disk.

The problem, and it's a big 'un, is that it doesn't work. I turn the disk and get nothing but a distorted scraping sound from the horn. No nuance over the circumference of the disk, not even a hint of recognisable speech. Pah.


So why doesn't it work? I'm not sure but perhaps the application of some physics might shed some light…

In its current form the read head has to move up and down a maximum of roughly 1.5mm. The head is pushed up the waveform hills by the contours of the disk and falls back into the the next valley through the power of gravity. I began to wonder just how fast the read head could move using gravity as its power source. Here is the equation used to work out how long it takes to fall a given distance in any particular gravity field.

t = √(2d/g) where t is time in seconds, d is distance fallen and g is acceleration due to gravity.

To save the effort of having to actually do any calculation I found this wonderful website:

www.gravitycomp.com which works out all sorts of gravity related calculations anywhere from the Sun to Neptune with, of course, a section for planet Earth. Isn't the internet a marvel!

The read head falls a maximum of approx. 1.5mm, plugging in this information into the appropriate box on gravitycomp.com  gives a fall time of 0.0175 secs. It sounds fast, but is actually a problem.

Double that time for a full up/down cycle giving a total time of 0.035 seconds for a full cycle which means that the fastest possible vibration speed of the read head is just over 26 cycles per second (26 Hz) This is at the low end of human hearing and certainly too deep to be useful in the playback I was hoping to create.


Where next?

I'm not sure. Talking to my Dad about it he pointed out that the groove in vinyl records move side to side, not up and down so gravity doesn't come into it.  Also they are also considerably smaller than the paper wave form that I made. I found this picture on www.synthgear.com.  Using the scale at the bottom of the

electron-micrograph it looks like the maximum movement is around thirty micrometers, a factor of fifty times smaller and well out of the range of cutting out with scissors. I don't want to accept defeat yet. It would be really cool if I could get an entirely paper playback device working so I'll be spending some time thinking about and researching this. It the meantime, any ideas? Any suggestions?

Comments (7)

  • Shelley Noble August 1, 2013 at 9:43 pm

    You will get this brilliant

    You will get this brilliant idea to work. It will likely end up being some kind of wave form picked up by some kind of sensitive plastic tag or some such for greater fidelity than the card could manage. Or maybe a sort of needle running over some sort of embedded grooves. I don't know, physics is way beyond me obviously. But I know you are creative enough to sort it.

  • Rodrigo August 2, 2013 at 1:16 am

    My analysis here is based on

    My analysis here is based on my physics hunch, far from technical.

     

    I don't mean to be a joy killer, but I don't think it's gonna work. The problem I see is that you are using a representation of the sound – the geometry of the wave in the software – and trying to play it through a totally different process.

    When a vinyl disc – or anything like – is made, there is a direct link between the sound and its representation in the grooves. The grooves themselves can only be used to play the original sound if a similar process – only reversed – is used.

    Same thing goes for the computer. The recorded sound have a digital representation in the computer. That representation can only be used to reproduce the sound in a similar way. If you use anything incompatible with the standard – software or hardware – you will get no result.

    What you did was to take one form of representation – the graphical computer wave that, in itself, is yet another layer of representation of the digital sound, a visual one – and turn it into something like the groove of the vinyl disc.

    Of course the vinyl disc and the graphical wave have some things in common, like the ups and downs due to "strength" of sound. But I don't think that is enough to reproduce the sound recorded in one by using the other.

    As I said, this is just a hunch. I may be wrong, but I'd suggest you to check the technical part, if possible with some expert.

     

    In any case, good luck!

    • Smelter August 2, 2013 at 8:04 am

      Looks to me like you need to

      Looks to me like you need to keep the "needle" hard up to the edge.
      How about taking gravity out of the equation and have the disc horizontal, then you could use a paper spring, maybe a single arc of paper to keep it in contact.
      Rodrigo may be a "kill joy" but he does have a point. Even vinyl records have a limited bandwidth which is why the audio is passed through an "RIAA" filter. http://en.wikipedia.org/wiki/RIAA_equalization
      You will probably have to enhance the shape of the wave form and apply pressure to the needle to get nearer to something like a real sound.
      Best of luck Rob.
      • JayHavvic August 2, 2013 at 11:22 pm

        There may be hope.
        It took a

        There may be hope.

        It took a little time to find, but Thomas Edison's recordings used Verticle or "Hill-and-dale" recording. The needle moved up and down.

        http://en.wikipedia.org/wiki/Edison_Disc_Record

        It does say that one of the down falls of the method was low volumes.

        Maybe part of the problem lies with the material used for the diaphragm the "needle" is attached to?

        Try, try again!

        • robives August 4, 2013 at 8:10 am

          Thanks for your feedback

          Thanks for your feedback everyone, I have a few more ideas to try and will be reporting progess on the blog. It'll be a few days though as we are having a family break in the Isle of Skye 🙂

  • cwolt August 9, 2013 at 1:31 am

    What an intriguing and

    What an intriguing and audacious project!  It's certainly gotten me thinking, and I think it would be wonderful if you could get it to work.  Though, the more I think about it, the more challenging it seems.

    Here are some thoughts:

    First, I would guess that the scraping sound you get is the result of your device picking up, and turning into sound, the roughness of the paper itself–all the little fibers on the surface of the paper.  As I was thinking about that it occurred to me that (of course!) the things used for phonograph-type recordings–vinyl, and in early versions, wax and metal foil–are all very smooth, so that the needle would pick up the wiggles in the groove, and not any roughness in the surface.  So one challenge in this project would be to get the surface of the "recording" very smooth. Would a thin coating of glue over the whole surface work?  Or a layer of wax? 

    I think the biggest challenge, though, is what you identified, namely that the "read head" has to be able to move very quickly.  It has to move back and forth probably thousands of times a second, in fact.  Here's a brief article: http://en.wikipedia.org/wiki/Voice_frequency that states that intelligible speech generally uses frequencies up to 3400 Hz, that is, 3400 back-and-forth vibrations per second. 

    I can see two challenges:  first, the "recording," which you're cutting by hand with a scissors, has lots and lots of little squiggles even if it's just a second long. (I think you've already taken on that challenge!) Second, the read head has to be able to move back and forth thousands of times a second.  That becomes easier to do the smaller and lighter it is, and the smaller the movements it has to make.  Making it smaller and lighter might be relatively easy to accomplish, but to make the movements smaller means making the contours of the recording flatter and more subtle–and far more challenging to cut accurately.  And I would guess that the subtleties of the waveforms are actually pretty important, especially for speech; I think it's the particular mix of frequencies, of tones and overtones, that we hear as the difference between different speech sounds, between, say, an 'a' and an 'e' and an 'h'.

    Here's one (probably crazy and impractical!) idea, of how to avoid having to cut tiny but very precise wiggles, and yet still have the read head make very small movements.  I'm not sure if I can make this intelligible, but here goes: imagine, first, rotating the read head  90 degrees from the position you have it in, so that it is now riding on the surface of the disk, the way the needle of a phonograph does.  Of course, in that position it won't pick up any of the wiggles in the edge of the disk.  But then, imagine rotating it back just a bit towards its original position; the read head will begin to be moved just a bit by the wiggles.  Even if the wiggles you've cut are relatively large, the movements of the read head will still be quite small.  If you need more movement, you can rotate it a bit more.  So, by adjusting the angle of the head, you can adjust how much movement it is required to make. 

    I hope that makes some sense.  I can think of lots of reasons why that might not work . . . but I'm intrigued by the possibility.

    I hope those are some useful thoughts.  I'd be fascinated to see if you can make this work!   And even if you aren't able to get it to say "Hello" to you, I wonder if it could still make some interesting sounds!

    That’s an intriguing idea! Using a sort of vector to cut down the movement, it is certainly worth a try! – RI

  • Lieven August 9, 2013 at 10:36 am

    How about printing out the

    How about printing out the waveform and cutting the paper in two along the waveform. Then you glue these two pieces on the side of a cilinder, and you take them apart half a mm. That way you create a groove for your paper needle to move in.

    Also I would make a sharper paper needle, maybe a sharp fold, and/or (not the goats) make the wave form larger.

    Great project!

    Lieven

    Another fascinating idea. It might actually be easier to cut as I could use a knife in roughly a straight line. Well worth a try thanks – RI

Comments are closed.