Open for suggestions on this one....
Most here have the same opinion of the Ovilus as I, it lacks scientific basis for use as a communication tool. An argument I often use is to demonstrate the mechanics of how the "spirit" is able to manipulate the device to get it to speak the desired phrase. To date, no one has done so.
I go into a similar discussion recently and we put our heads together and attempted to determine a means to allow for this interaction. I had done and still have the test equipment used to develop speech using a synthesizer from back in the 1980s. (General Instruments SPO256) For my purpose I built a keyboard which allowed me to enter the phoneme command for any of the 64 phonemes this processor utilizes. The Ovilus also has a phoneme mode but lacks a means of directly addressing the phonemes. So the obvious solution would be to provide such an interface.
I looked over the available phonemes used with the SP0256 and found several were similar. I found that to simplify things I could reduce the number of phonemes needed to 30 and still get reasonably good speech. All that would be needed would be a type of keyboard that in theory an entity could interact with.
What I came up with was a large panel with 30 "keys" spaced about 4 inches apart. The letters, A - Z would go from left to right and be identified as such. Some letters, especially the vowels, would have 2 or 3 "keys". One for a long sound another directly below it for a short sound. Other letters would also have dual keys, such as the "C" since "C" can take on either an "S" sound or a "K" depending on the word desired. A few letters such as "Q" would actually not have an input, rather the "Q" would be tied in with "K" since phonetically that closely matches its sound. I feel this would be sufficient to allow for very basic words to be formed even though a few may take on a bit of an unnatural sound at times.
The keys themselves are another factor. Each key consists of a Field effect transistor with an open Gate. This makes it susceptible to any static field that might interact with it. There would also be a mechanical open contact switch that would permit a physical action to close the key. Additional types of "keys" could also be implemented if later versions are built. Each key is also surrounded by a solid wire that is connected to a pause command. This would be detected should an interacting field simply pass directly between keys without turning off. Its physical location would be "mapped" by adjacent keys activated. Later analysis might allow for "correcting" any such errors in event of multiple key activations.
The output of the Phoneme Keyboard is sent to a converter which does two things. First it converts the key activation to the proper phoneme address needed to cause the SP0256 to speak that phoneme. Secondly it permits one and only one activation for a given key preventing multiple hits on a single activation. And for any who may wonder why I am building this for an old technology speech synthesizer, I have a working processor already going. I would have to do minimal software work to adapt to this. It is much easier in my case. I would expect that newer technology, Arduino or USB for example, might also be utilized. Changes to the converter would be required since as I built the original processor it requires a parallel data bus.
This is a brief outline of the project as I currently have it. I am open to additional comments and ideas to improve or enhance its ability.