Theremin & Talking Machine

@ coalport....I saw your recent Youtube clip with the Goldberg theremin and the talking machine from GordonC's FaceBook newsfeed...Also posted by Lydia Kavina on her feed...My jaw is still on the floor....Fantastic...Gordon suggested that you had somehow stuffed Pavarotti (sp?) into the cabinet...It really did sound like that...I also didn't notice that the lightning bolt pitch antenna caused you any issues in playing accurately...Bravo!  jcn

This "talking machine" really does a believable vocal emulation, really impressive, particularly in the hands of a master! I had heard this machine with guitar and it was impressive - but these 'performances' by Peter, IMO, are astounding.

I wonder if anyone knows how many formants are being filtered ? - I have played a lot with analogue formant synthesis - working on having the volume antenna as a 'virtual joystick' with 'X' position controlling the first formant and 'Y' controlling the 2nd formant (Z controlling volume) so that one could get the theremin to 'sing' under ones manual control.. 2 formants are not really believable ( I was trying to add more formats linked to X and Y co-ordinates, but had almost given up on making it believable, and was thinking about it more as a vocal 'tone' effect) ... What I produced never came close to what this machine is producing! (in terms of realism - but manual control of the formants does make it possible to control what is being 'sung')

If it was analogue, Id buy one to see how it worked - Alas, its probably a DSP and everything is locked away in its digital heart! -

Yup, there's a 2cm square Analog Devices chip with gazillions of legs centre stage, with a smaller rectangular EtronTech chip next to it, an lots of little chips and resistors and capacitors surrounding them, all surface mount.

I also see a ten pin male connector, I guess a serial port?

Not surprising its an AD part - almost certainly one of their wonderful DSP's (the only DSP's I have ever played with are AD - they do low cost development boards and software) .. The 10 pin connector could be anything, but probably includes SPI or similar serial interface allowing programming of the part - other pins could be test points or whatever.

Reverse engineering is likely impossible - code in the DSP will be read protected (unless the manufacturer made a huge error) and even if not, pulling out the code and decoding it and doing everything else (deriving schematic etc) needed would take skilled man months - Someone capable of doing this would probably be able to design a TM quicker.

All I want is a chip which takes an 'x' and 'y' signal and adjusts the formant filters to plot this co-ordinate onto the 'map' .. I am (almost) sure that with a little modification, the TM DSP could be made to do this.. but this 'modification' is something only the manufacturer would be able to implement.

Fred.

Gordon! You opened your Talking Machine and had a look inside. Were you able to get the board out of the box? I removed all the anchors that are on the outside but I still couldn't lift the board. 

I am convinced there's a little man in there somewhere.

The real test would be to feed a few seconds of white noise through this baby and post an MP3 of that.  Peter (or any other owner of the TM here) could you perhaps oblige us?  If so, I would be happy to post an analysis picture or two of it.

Peter! I didn't try. I didn't want to startle the little man.

"I am convinced there's a little man in there somewhere."

If you find a little boat amongst the ICs and such you might be getting warm. ;)

i think this is the best of PP so far:

http://www.ehx.com/blog/ehx-stereo-talking-machine-and-theremin

"i think this is the best of PP so far:

 http://www.ehx.com/blog/ehx-stereo-talking-machine-and-theremin"

I agree - absolutely incredible! 

Peter, could you describe what kind of tone spacing are you using?  It looks pretty wide.