I have the electroglottograph circuit prototyped on a breadboard. It works - you can clearly see the signal, but it's very noisy. And that's only when a man's vocal cords are being probed - the signal taken from a woman's vocal cords is much weaker (probably due to the larynx being smaller in women), and it's almost completely hidden by noise.

The circuit shown on the project's page is actually one of the later iterations of the project. Previous versions included:

• A simpler oscillator, based on a CMOS 555, instead of the Colpitts circuit with a ceramic resonator currently used;
• A decoupled voltage divider instead of the TL431 for the mid-supply voltage;
• A rectifier-based AM detector (using two op-amps and two diodes).

All these changes did not do much to reduce the noise. Where does it come from? I don't know. Does it come from the supply (I currently use a 9v battery regulated by a 7805)? Maybe, but in this case better decoupling in key places should make some difference, but it does not. Is it radio noise? Maybe, but the cable used for the probes is already shielded and twisted, and I tried shielding the entire prototype circuit; no real difference. Is the voltage-to-current converter this noisy? I'm not an expert, but I tried calculating the theoretical noise level, and it's lower than what I observe.

I'm currently abroad and do not have access to my scope, so I cannot look for the noise more directly. I also do not have soldering equipment, so I currently can't check if this is just a breadboard issue. If somebody has more experience building these kinds of circuits and has some idea about the possible source of the noise, I would gladly hear about it.