Speech Coding .001

Could a coder write real software using just speech? Of course, a coder could use dictation to enter lines of code into an IDE and perhaps nudge and edit here and there, and then compile to product an actual, entirely original compiled binary application that was coded using speech, but that isn't what I mean.

I also don't mean having some application that can be manipulated or configured using your voice, because like the quintessential "Microsoft Office Programmer" or "VCR Programmer" that isn't really programming, it's just configurating.

Could a coder write a real piece software with speech and only speech -- that is to say that there is no visual development environment, and no visual feedback. Only the coder talking and the development environment speaking back or making sounds make up the programming and debugging environment.

If it sounds impossible to you, you are mostly right so far, except in the most narrow of niche apps where the application itself is nothing but speech and speaking.

A small IVR company in Tampa Bay is experimenting with an IVR scripting module that allows the IVR programmer (OK, maybe a glorified IVR configuratior) to program the IVR using only voice and audio feedback.

Although the process is kind of like playing checkers while blindfolded, it does demonstrate a very primitive example of how a person and a server could interact to produce a runnable piece of software entirely and exclusively by voice.

In this audio transcript, the programmer records a prompt with two options, gives it a name, then reviews the available prompts ("List prompts") and then produces a "main" procedure line by line. "Play greeting", "Get Digits", "Store in X" which then proceeds to submenus for digits 1 and 2.

A small, mostly useless but functional IVR menu and program flow is produced and ready to run in about a minute.

http://www.GotSomeBeats.com/speechcoding.mp3

It might seem somewhat tedious, but with a fair amount of concentration, practice, and with eyes squeezed shut a programmer could produce fully functional and quite extensive IVR interactions in a fraction of the amount of time it might take using conventional IVR programming methods. Additionally, changes to the IVR could be made by voice over the phone by the programmer, shared library routines could be imported by voice request, and toolkits such as payment processing procedures, address and identity verification, voicemail assistants, etc, could all begin working together in a mysterious cloud of pure audio.