Speech synthesis has come on a long way since the days of this device :)

Feedback sent to Cepstral, and their amazing voice lawrence :

How much time and labour does it take to create a new voice roughly ?

What price would you be talking, to generate a voice like Lawrence ?

I think lawrence is very good but still has a little non neutral bias in there, if you ever come to create a truly neutral clear english voice, I’d be interested to try out. Also strange in testing the demo, that millie is capable of pronouncing the word “barn” but lawrence is not.

These pronunciation errata can obviously be corrected through misspelling the word until the speech voice gets it right, I guess a way of leveraging these pronunciation corrections, would be a pronunciation correction interface bring-able up on screen by hotkey, that would in realtime send a pronunciation corrective word translation back to a central database at cepstral to then be shared via update with all users ? such that all share the benefits of community labour towards improving the pronunciation dictionary of the voices.

Great work why do Apple not licence the voices from you ? As the system speech synthesis voices have not reached the phase of being bearable to listen to, in comparison to the Audio version of a book say. And also I dont think speech synthesis technology can ever reach its potential unlees the best quality voices are guarantee-ably usable from with any of the mainstream operating systems. Much like say Helvetica as a font, can be relied upon to be available on almost any computer.

When like photo-real photoshop composites, people cannot tell the difference between a voice artist reading the book in audio form and the speech synthesis voice of that same artist reading it, then I guess speech synthesis will truly have arrived as a medium. this ideal is probably like chasing a rainbow in terms of complexity involved with speech timing patterns but hopefully one day. Good luck with your endeavour’s.