Lazyweb, I need text-to-speech into the voices used to announce BART trains, which are Lucent / Bell Labs "John" and "Grace", from the mid 1990s. They sound like this. This is for purposes of Important Comedy.If you can't find either of those precise voices, can you find something pretty close?
Maybe this? http://real-votrax.no-ip.org/
I know they say Lucent but it really sounds like DECTALK to me. Here it is for MAME, works for me: https://www.emuparadise.me/M.A.M.E._-_Multiple_Arcade_Machine_Emulator_ROMs/DECtalk_DTC-01/179357-download
Oh boy. I haven't tried to run MAME since like... 2003, and it was a monumentally unpleasant experience. What do I need to download to actually do something with that? MacOS.
I was able to get it to run by saying brew install mame but YMMV. The interface is still atrocious; I thought it had hard-locked my Mac (there's an experience I haven't had in 20 years) but instead it was just fullscreened on the desktop without giving me any way to get there (so only launch it from a windowed terminal).
Running the DECTALK ROM took minimal hoop-jumping. It sounds just like Stephen Hawking (so, you know, that's cool), but not "John" and "Grace".
I was able to get Mame on OS X to work after jumping through several hoops documented here:
https://www.evernote.com/l/AAHSmHVFQdNML5h8FkV9lweJfcLJoeGea38
The dectalk rom worked even after mame complained about the machine being broken. A fun exercise, but won't result in John and Grace.
You want SDLMAME, it's a standard 64-bit macos command line app.
https://sdlmame.lngn.net
Pretty sure it's Lucent. Sadly all their online web stuff is no longer functioning, but some of the recordings are archived:
http://web.archive.org/web/20060207133442/http://www.bell-labs.com/project/tts/index.html
The fossil one really tells me that yes, it's the Lucent/Bell one and not DECTALK.
Also.... http://web.archive.org/web/20060211130703/http://www.bell-labs.com/project/tts/daisy.wav
1) Definitely not DECtalk. Have a friend in college who used a DECtalk with his laptop. Different voice. It is Bell Labs.
2) Until the internet bitrotted away they had a functional site that would have let George and Gracie say whatever you want, but it's dead now. (Thanks, Nokia!) But it has a contact for inquiries, Michael Tanenblatt, and if he is a True Geek he would be happy to hear from you. Here's his (ugh) LinkedIn: https://www.linkedin.com/in/tanenblatt/
which are Lucent / Bell Labs "John" and "Grace",
John and Grace? They're George and Gracie, you uncultured lout! Though the computer voices sound like whales, they even did the bit!
Not according to the linked article. John and Grace were the original voice names. BART renamed them.
Very well, I should've read the article instead of just listening to the recording (where they introduce themselves under their BART names). I retract my comment.
Cool story, bro. Thanks for showing up.
I think this is the modern evolution of the Labs TTS SDK, but they seem to have dropped or renamed George & Gracie
http://www.wizzardsoftware.com/text-to-speech-sdk.php
Wayback has that page from 2012:
http://web.archive.org/web/20120227142711/http://www.wizzardsoftware.com/text-to-speech-sdk.php
It has a much longer list of voices than today's page. But no George & Gracie.
Sounds like the output from the federal screw works voice synthesis chip. I think they went defunct in the early 80s but maybe someone bought their IP and sells a similar part.
I just tried to install https://www.plogue.com/products/chipspeech/ but some failed Visual C++ dependency kept it from working on my version of windows.
Bell Labs had created SABLE which was an XML markup language for synthesis. Perhaps Festival (which has SABLE support) might lead to clues? Clearly Bell Labs/Lucent collaborated with CMU (http://www.speech.cs.cmu.edu/)
https://www.cs.cmu.edu/~awb/festival_demos/sable.html
http://www.cs.cmu.edu/~awb/festival_demos/general.html
Lastly a demo of "Festival Lite":
http://tts.speech.cs.cmu.edu:8082/