Text-to-voice programs are becoming better and better at reproducing the sounds of natural speech. Their voices have almost lost that boring robo monotony they used to have and they’ve become a bit more responsive to the clues that punctuation offers about rhythms. I thought it might be fun to see — hear! — how an entry in Slaw would sound if read, well, automatically.
Because I wanted to try out a number of different voices, I used the free demo function of Cepstral Text-to-Speech [all platforms supported], rather than buying (at US$30 a pop) any particular “speaker” just now. That meant I was limited to only a few words, which is fine because it means smaller MP3 files for you to download to listen to. The English excerpt is taken from Agnese Caruso’s recent post:
“Becoming smarter about new sources of information” is one of those skills. Since today’s students are surrounded by blogs,
And the French excerpt from Michel-Adrien Sheppard’s post about Montreal prosecutors:
Le responsable de la sécurité des procureurs au ministère de la Justice, Me Sabin Ouellet, reconnaît le problème. «Bien oui,
I’ve done three voices: David, who speaks US English, Lawrence with his British English, and Jean-Pierre speaking “Canadian French” (and reading a French excerpt, of course). I think they’re not bad. (Notice, though, that Jean-Pierre doesn’t recognize the abbreviation for Madame.) Cepstral offers quite a number of different voices, women’s as well as men’s; and you have the ability to raise or lower the pitch and increase or decrease the speed. I’ve chosen three men’s voices to enable you to compare them better.
The short excerpts of text don’t really let them strut their stuff across the full minefield (wow, there’s a dangerous mix of metaphors) of legal prose, with its citations and long subordinate clauses, of course. But I think that they might let you listen to Slaw without causing you to go mad; and they might let someone turn a print blog into a podcasting blog without a whole lot of difficulty.
Let me know what you think. Talk to me. Here’s talking at you: