mersenneforum.org  

Go Back   mersenneforum.org > Extra Stuff > Science & Technology

Reply
 
Thread Tools
Old 2015-08-20, 08:31   #1
xilman
Bamboozled!
 
xilman's Avatar
 
"π’‰Ίπ’ŒŒπ’‡·π’†·π’€­"
May 2003
Down not across

72·11·19 Posts
Default Build your own Dalek.

Intel has made Stephen Hawkins' speech synthesis source code vailable on GitHub. Windoze only for the moment.
xilman is offline   Reply With Quote
Old 2015-08-20, 10:55   #2
LaurV
Romulan Interpreter
 
LaurV's Avatar
 
Jun 2011
Thailand

2×11×397 Posts
Default

Quote:
Originally Posted by xilman View Post
Intel has made Stephen Hawkins' speech synthesis source code vailable on GitHub. Windoze only for the moment.
I designed and build the "Chessvoice" (for European chess federation), some time ago. The device is for blind people playing chess in official championships, and it has a pair of headphones and 3 buttons, you put the headphones on, press this button or that, the device is "watching" the official chess clock (the one which records the time and both players are pressing the buttons) and it will tell you the time in the headphones (so you don't disturb the opponent ant the other players). Your time, opponent's time, number of moves, etc, according with the button(s) you press. It is kind of "text to speech" machine, and all the "phonemes" are stored as 5 bits wav files (extremely small wav's in memory, all the memory of the MCU is 32KB only, I used an ATmega, 8 bit Atmel CPU at that time). What is nice about that device, I use a neural network (SOM like) to combine the phonemes and the tones, for example the 5 in "5 minutes" or "25 minutes" or etc. are all pronounced different, in such a way that it sounds REALISTIC. You don't know that a machine is speaking. You also can select the timbre of the voice, like in "male", "female", child" (and I only recorded the voice for "child", is my daughter's voice, who speaks as a native speaker). You imagine the "lady" says "you have one hour, twenty-five minutes and sixteen seconds left, black is to move", and this sounds EXTREMELY realistic, not like when you dial the number to hear your bill at your mobile and you hear something like "you HAVE five, dollars; and TWENTY NINE cents left AND your subscription will. Expire on TWENTY and fifth OF September". If you know what I mean. Like in the train station too.. In fact, I don't remember in my life to have been listening to a clear and realistic text-to-speech, EVER, except for this device. Well, the advantage is that I had only a limited number of words, mostly numbers up to 59, few fixed phrases, to implement there, but I challenge you (general you) to store a single wave in so less memory, without talking about so many other waves, or executable code (they must all fit inside). (That is why I had to cut the wave, to let it on 5 bits instead of 8, as it was recorded, I recorded the girl saying all numbers to 59 few times and all the phrases, all over few MB of data, then I took Audacity and start splitting them in phonemes, as short as I could make them to be still possible to be combined in words). I could feel for at least 5 patents with that product, including for data compression (I never filled for a patent in my life, I wouldn't know how, but some of our customers have patents on things I "invented", or developed, sometime putting them in their products even without their knowledge, they found out later when I explained how I did it... Then they applied for a patent and got all the rights, without asking me/us - of course they paid the development of the products, but well... )

Last fiddled with by LaurV on 2015-08-20 at 11:03
LaurV is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
New build help Prime95 Hardware 147 2018-11-10 00:58
Weird build? Xyzzy Hardware 122 2016-01-14 10:32
New build with ECC ram PageFault Hardware 1 2012-08-23 03:44
New build PageFault Hardware 28 2011-09-19 07:15
Hypothetical Build Kevin Hardware 4 2009-05-27 14:46

All times are UTC. The time now is 14:34.

Fri Sep 18 14:34:01 UTC 2020 up 8 days, 11:44, 1 user, load averages: 1.50, 1.52, 1.46

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.