Thanxx, MOSC...
Yea...i'm new 2 the world of the G2, but not 2 synthesizers in general...
Made some patches in the month that i have this wonderfull machine...but not like most of you guys are programming...
Thanxx for the patches guys !!!
Mass Destruction - A G2 performance which is a work in progress. Demonstrates the G2 as a speech synthesizer among other things. This is a real-time recording.
Very cool. How did you do the sequencer module programming to get those words out of the vocoder?
I started with a wav file generated from the excellent online TTS demo from AT&T. I then used Praat www.praat.org to get the frequency curves of the 1st three formants. The sequencers control the frequency and amplitude of three sine oscillators. The sines are just mixed and fed to the vocoder. The 1st sequencer row also switches between voiced/unvoiced modes. Praat is an excellent application for studying speech. I started out closely matching each point one at a time but found that a visual approximation of the curves was close enough. The phrase is broken down into four sections and sequenced by changing variations.
If I can figure the internal structure and checksum calculations of a patch file, I may write a program to automate the translation from Praat to a generic talking pch.
For those who are curious, if you patch the summed sine waves directly to the audio out, you can still understand the speech (so long as you know what it is supposed to be saying). See also http://www.haskins.yale.edu/haskins/MISC/SWS/SENTENCE/S9.html
That is a facinating web site. Listening to those sine wave files is amazing. Thanks. I've studies speech and audio for years and never heard these kinds of tracks. No matter how smart I think I am, it never surprises me to find out I'm not. _________________ --Howard
my music and other stuff
I feel a bit stupid as well now, I'd been reading about the praat software in a newspaper a few years ago and how amazing it was, thought it to be interesting and even dowwnloaded some stuff. But I failed to see the light ... I'll have another look at it.
Sofar I only managed to listen to the mp3, but as a speech example I think the patch is very good.
Also I thought it to be interesting to note that Kees van der Maarel has been making several speech patches (for G2 but also for NM Classic) in the past by using a diferent (maybe less methodological) approach, but also with amazing results.
Sort of finished now. Not too different from the original. I had planned on lots of Mass Destruction sounds but when I looked for them, I couldn’t find any.
The speech block is simplified, turns out that the separate volume curve for each formant is not strictly necessary.
Ian, this is the coolest use of automated variations I've seen yet. Get's me thinking about making more complex state machines without uneccessary modules in a patch.
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum You cannot attach files in this forum You can download files in this forum
Forum index » Clavia Nord Modular » G2 Performances - Experimental
FAQ
Calendar
Search
Memberlist
Usergroups
Register
Profile
Log in to check your private messages
Log in
Chat Room
PsYx, glad to have you here...
