GetLoFi has found a real gem in the form of the Flame talking MIDI synth. Hacking Speak ‘n Spell’s? Totally last year. This synth is homebrewed from the ground up, using an IC (integrated circuit) that specializes in speech synthesis. GetLoFi says that chip is permanently on backorder … why am I not surprised? The chip offers “2 speech elements (allophones), 43 sound effects, and 12 DTM touch tones.”

Flame-The MIDI Talking Synth [GetLoFi]
MIDI Talking Synth [Analogue Haven sale page]
Flame on Analog Damage months ago, while I wasn’t paying attention (sorry, Chris — still interested in tips on using this IC, though!)

There’s a microcontroller in there, as well, translating joystick and MIDI inputs to serial for the speech ICs (dual SpeakJets, in this case). This is already a great achievement. But is anyone else imagining an open sourced version of this, particularly given the possibility of features?

And might the current circuit bending trend ultimately give way to a bigger “build your own music hardware” trend? Part of what I admire about bending is its interest in chance and re-appropriating existing hardware. But I see potential for the growth of instrument construction culture, as well.

Dave's Imaginary Sound Space has an incredible compendium of links on the history of speech synthesis, complete with sound files, all the way back 50 years to the Sonovox, connected to a Masters' Thesis by Sami Lemmety.

Inspired by Dave's links, CDM has decided to bring you more of the time-sucking power of fascinating speech synthesis links!

Len Sasso has a great article from the 2/2004 Electronic Musician with everything you wanted to know about speech synthesis — history, how to do it, and more copious links!

You can hear how the National Weather Service (USA) has changed its voices
– for those of you who regularly listen to these broadcasts, you can
wax nostalgic about your old friend Paul, who just wasn't as
cutting-edge as the new-fangled Craig and Donna.

And lastly, for the best-sounding speech synthesis I've heard, listen to the eerily-accurate sounds of AT&T's TTS
multi-lingual, interactive demo. Enter text, hear it, and even save as
a WAV. (I want to hear the songs you create with it, okay?) And yes,
you can still do what Radiohead and others have done and use the Mac's speech synthesis capabilities.