This is at the core of my current effort, which BTW has nothing to do with replacing EVA, and everything to do with providing myself with a transcription that encodes as much information as possible, so that each time my views change somewhat, I don't have to go back and redo my transcription. If enough information were stored in any current transcription, this would be unnecessary, as simple global changes would suffice in all but a few cases.
On the good side, once this transcription has been offered, argued out and proofed by all interested parties, it's a simple matter (as Rene points out) to port it to an extremely accurate EVA transcription.