Sing-Song Speech

Before I lose my train of thought, for text-to-speech, the mannerisms for speaking might be best serviced with a music generation system.  My reasoning behind this would be it is not necessarily music but it has a certain meter, pitch variation and so on.  If it were to come down to an AI coming up with something to say, mood can best be expressed by a variation in tone in much the way as something is sing-song.  However, a certain clamp would be implied so that it isn't over the top, but at the same time breathe a little bit of life into the character.

I had something else but honestly couldn't tell you where I was going with the thought.

Powered by Qumana

Missed a blog

Already I missed an appointed posting. Need to keep up with it.

As far as last night goes, I've been bouncing between WoW, Naruto X-Box, keeping up with emails, and daydreaming. For WoW, I've been working on my Hodir reputation -- which is the same 4 tedious quests with a mild inuendo to the titles. For the Naruto game, I'm amused with the story mode that they have -- tho the best part of the game is the one where you jump through hoops in the forest competing for best time. So fun. :D

Lastly, e-mails and daydreaming. I keep thinking to myself if I spend the money and time to build a cluster/cloud of VMs with a rackmount that I can somehow get myself back into programming and make a Neural Network / XML distributed DB. Speculations on what it should have as a primary task I haven't an inkling of what to do with it. Perhaps have a way to store character information for an unnamed Wayfarer MMO (or even just a paid content site), or something with Shadowrun, or miscellaneous -- maybe a CMS/Bulletin Board/Fuck-all-if-I-know.

We'll see

Hooray for Idle Time

While I was waiting, I managed to get bugzilla, tikiwiki, and subversion installed on the new VM I made.  So close to getting a more secure, less hardware-dependant server up and running.  Once I'm fine with everything going up, I can switch the DMZ and point it to the VM instead.

Powered by Qumana

Tags: Servers, Web, Subversion, Virtual Machines

Circuitry and Artficial Life

So yesterday I had a few thoughts on rolling my own circuit board for a cellphone.  There were quite a few useful sites on finding manufacturers for LCD screens and the like, but I'm not really sure on how to get things like an nVidia chip (just a chip with unsoldered pinouts), or an AMD subproc.  While I was digging around, I thought it would be neat to have a GSOM (Graphical Self-Organising Map), first to place the chips using heat-distribution rules (probably some complicated math/physics formula), and then treat electrons like ants.

The ants would have their own rules to follow:

• Shortest path from 1 pin to another


• Minimum distance from another ant's trail


• Obey things like resonance, interference


• Making "fuel stops" at capacitors, LEDs, so on


• Possibly re-routing 3-dimensionally up to 8 layers

I've always wanted to have a program that writes programs, what about a program that makes hardware -- but intelligently.  Another factor is that there is debugging tools out there to run mockups of the circuitboard before going to fab, I just wonder if there was a way to interactively debug then let the GSOM program adjust and learn from it's mistakes.

Food for thought at least.

Powered by Qumana

Tags: SOM, AI, Programming, Hardware, Random Thoughts

Offline Blog poster thingy

So we'll see how Qumana works better than w.Blogger.  I suppose first thing would be that w.Blogger requires that you have some knowledge of HTML tags -- in addition to not having a very WYSIWYG interface:

• Doesn't *actually* post, nullReferencePointer (meaning something got upgraded


• Has been out of date for 3 years running


• Just not fancy enough.

Although really, if only Qumana could be skinned a lil better, as well as the option for plug-ins.  I'd like to show the playlist that I've been listening to and so on.

AI Programming

I've been eyeballing a few things to code with, tho trying to figure out which one and why I would want to use it. For instance, should I go with LISP of which is really good for lists handling weighted responses or something like F# which seems to integrate well with any other .NET assembly? I dunno. Things like a neural network seem to be pretty dependant on having nodes of some sort, with weighted values on the connectors.

??:/??

Yeah, I guess I need to brush up on my math skillz. I've been needing the brain exercise in a long time -- it seems to have atrophied to the point where my memory sucks and doing simple addition takes longer than I had planned.

LISP, C#, and MIDI

So there are a few things of note (haha) that I have been looking up. For starters, I'll begin with XML for MIDI as well as tones and the like. On wikipedia I've found a piano scale that has a listing of the mhz in a chart, so being able to define (for instance) a song definition to convert over to the MIDI format would be a lil' bit easier. We'll see tho.

Next on the agenda, is actually getting down the byte-code format for MIDI itself. I'm totally flabberghasted that there isn't an already pre-built opensource creator -- but I see that as a challenge to rolling my own. I'd probably want to set up an SVN server at home just to sync it up between work and elsewhere. That is going to be a pain just due to the fact that I can't even get samba working.

>:(

Having alotta prerequisites does not make for getting a project done. Maybe I should just git 'er done and not fret the speedbumps.

Meh.

On the topic of the MIDI generation, the other snag is LISP -- in particular, ANSI LISP. I don't know it, and sorta get the concepts. I'm guessing it is more optimised than an XML/bTree setup, since it has been around for quite a while and has a very specialised application for AI. Only thing I am curious about is in the SETQ the definitions are statically set (50/50, 34/33/33, etc.) and I want to have something akin to a neural network assign these values in a fuzzy manner -- such as floating points between 0 and 1.

I'm not even totally sure that a neural network is built to handle that sort of thinking anyhoo.

Something for later, I s'pose.

Oh right

So I haven't touched blogging in a while, however it is one of those things where I have:

• Last.FM: Tracks my music history and stuff
• Google Library: Tracks my book library
• Amazon Wishlist: Tracks the books that I want to get plus misc. doodads as well
• NewEgg Wishlist: Computer related wishlist
• Picasa: All of the pictures and photos that I uploaded
• Twitter: Quick microblogging-style posting system
• MySpace: Social Networking site (I mostly use it for the Games nowadays)
• Plus a few things that I can remember offhand

If only I had a system that would synchronise that crap together into one place. Maybe an app that I myself wrote just to handle all of these cases. Unfortunately, the stuff that I don't mess with as much doesn't have an API to begin with -- either because they want to keep their data to their own site (pshaw, to make money or some junk) or they just haven't gotten around to it.

Ah well, at least I can keep dreaming of some unified place to store this.

Aight, back to work!

EDIT: Oh, and completely unrelated -- I bring to you an image of Dizzy (from Guilty Gear X)