#47804 - headspin - Wed Jul 13, 2005 7:54 pm
DSSpeech Speech Synthesizer v1.5
(C) Copyright 1986 A.D.Beveridge & M.N.Day. All Rights Reserved
Windows Adaption by Stefan Stenzel
DS Port by Headspin (headkaze [at] gmail.com)
I would like to personally thank Stefan Stenzel for his most gratious generosity and kindness with providing the source, as well as taking much time to reply to my nurmerous and annoying e-mail's and nagging.
-Version 1.5 Added Keyboard (now it will say whatever you want)
http://headkaze.webpal.info/
_________________
Warhawk DS | Manic Miner: The Lost Levels | The Detective Game
Last edited by headspin on Sun Sep 11, 2005 9:29 pm; edited 1 time in total
#47805 - jstart - Wed Jul 13, 2005 7:58 pm
Cool! But yah not much purpose plus all that randomness made no sense whatsoever
But i guess that was the point
_________________
-=Jstart=-
#47808 - headspin - Wed Jul 13, 2005 8:17 pm
I uploaded the Win32 version here for people who dont have *me. You can type things for the synth to say. The DS version uses the same synth, so it would sound the same on DS.
_________________
Warhawk DS | Manic Miner: The Lost Levels | The Detective Game
#47813 - Extreme Coder - Wed Jul 13, 2005 9:03 pm
Looks and sounds good :)
So you're saying this is in the Neoflash Compo? Or you are entering with something else? I am entering too.It would be nice to know what are you entering with:P
#47817 - Veg - Wed Jul 13, 2005 9:44 pm
Heh, I love speech synthesizers. Reminds of the old Amiga workbench 'Say' Utility.
I remember using one... I think it was a 'Creative' program that came with a soundcard - it allowed you to set the pitch of each letter, so you could make the program 'sing' things.
Any chance you could add such a thing to this version?
#47837 - Diskun - Thu Jul 14, 2005 12:28 am
Impressive :)
If you discarded this for the Neoflash compo, what a wonderful thing you'll be presenting for it?
Greets.
#47875 - headspin - Thu Jul 14, 2005 6:57 am
Yeah, I do have something better for the NeoFlash comp. But I doubt I will get a place since all these wicked emulators are always very popular. Good luck with your entry too though :)
_________________
Warhawk DS | Manic Miner: The Lost Levels | The Detective Game
#48941 - pissant - Sun Jul 24, 2005 2:21 pm
| headspin wrote: |
| I uploaded the Win32 version here for people who dont have *me. You can type things for the synth to say. The DS version uses the same synth, so it would sound the same on DS. |
Is the source code available for this speech app? I've been digging around google for the last few minutes but haven't turned up anything yet...
Thanks
#48945 - headspin - Sun Jul 24, 2005 3:38 pm
The source in not available on the net, sorry. Microsoft provides a much better speech synthesis/voice recognition API for the Windows platform. I recommend using the Microsoft Speech API (SAPI) that is documented at the Microsoft Agent SDK docs.
Do you mind me asking why your interest in the source?
_________________
Warhawk DS | Manic Miner: The Lost Levels | The Detective Game
#48948 - tepples - Sun Jul 24, 2005 5:08 pm
| headspin wrote: |
| The source in not available on the net, sorry. Microsoft provides a much better speech synthesis/voice recognition API for the Windows platform. |
That won't help if I'm trying to port speech synthesis to a platform other than Microsoft Windows.
_________________
-- Where is he?
-- Who?
-- You know, the human.
-- I think he moved to Tilwick.
#48954 - headspin - Sun Jul 24, 2005 6:16 pm
I havn't got permission to release the source from the original authors, so until I do that my hands are tied. If someone is genuine about doing a port or using the code for a specific project, then I will ask.
_________________
Warhawk DS | Manic Miner: The Lost Levels | The Detective Game
#48959 - tepples - Sun Jul 24, 2005 8:58 pm
| headspin wrote: |
| I havn't got permission to release the source from the original authors |
What reason did they state as to why they wouldn't let you publish the source code? Is it currently in use in a proprietary software product that is still in print? Or do they want something like the Blender3d bounty?
_________________
-- Where is he?
-- Who?
-- You know, the human.
-- I think he moved to Tilwick.
#48968 - headspin - Sun Jul 24, 2005 10:15 pm
There are two original authors of STSpeech, namely A.D.Beveridge & M.N.Day. It was ported from Atari ST 68k by a fellow named Stefan Stenzel. No ARM optimizations have been done yet, I really released it earlier than I wanted to and ran out of time to do any great improvements specifically for the DS.
Previous to this I had ported an old unix voice synth that uses phomeme contatination. The results are rather poor, but you can check it out yourself here and source is here.
As for the source to STSpeech/DSSpeech, you must understand permission must be granted by the original authors, even though Stefan practically re-wrote the synth engine. I do not know if he is still even in contact with them. I have posted Stefan an e-mail requesting the source go public, and we will have to wait for a reply. I'm quite happy to give it away, but you have to respect all that have contributed.
The source was not simply handed over to me by Stefan. I gave Stefan detailed correspondence regarding my intentions for the port. The problem here is commercial use of the code, which is against Stefan's wishes (I guess). He was very gratious to spend so much time e-mailing back and forth with a "homebrew" author. He wen't on holiday after sending me some of the code (parts were missing), he got back and was ill for a short time. I started to reverse engineer the binary to get important phoneme data structures! He came through eventually and sent the .def files. After a bit of tweaking I got it compiling in Visual Studio 6. Then I ported, quite painlessly, to the DS.
There is more to the story, but not really relevent to your question. But I was determined to get a friggen synth that didn't sound like Max Headroom on steroids. I think DSSpeech nearly gets there. But Stefan is the true genious behind this. I really should have a readme.txt with the ROM to explain all this!
_________________
Warhawk DS | Manic Miner: The Lost Levels | The Detective Game
#48975 - tepples - Sun Jul 24, 2005 11:18 pm
| headspin wrote: |
you must understand permission must be granted by the original authors, even though Stefan practically re-wrote the synth engine. I do not know if he is still even in contact with them.
[...]
The source was not simply handed over to me by Stefan. I gave Stefan detailed correspondence regarding my intentions for the port. The problem here is commercial use of the code, which is against Stefan's wishes (I guess). |
Point him to the FFTW project, which has released a commercial-quality Fast Fourier Transform library under the GNU General Public License. The copyright owner agreed to do this only because the copyleft aspect of the GPL allows the copyright owner to price-discriminate against those who rely on a typical proprietary software business model. The LZO data compression library, used by PocketNES for its save states and by early versions of Tetanus On Drugs for GBA for its assets, is under a similar license: GPL to those who haven't paid and proprietary to those who have.
_________________
-- Where is he?
-- Who?
-- You know, the human.
-- I think he moved to Tilwick.
#49231 - Chetic - Wed Jul 27, 2005 2:51 pm
This is awesome!
I remember playing with STSpeech for hours as a kid.
The sound is great, even on the DS :D
Aren't you going to add a keyboard to it so you can write your own stuff though?
_________________
Packin':
Grey DS with FlashMe v7
1Gbit XG2T 2005 (Neoflash compatible)
GBAMP, Supercard CF, 512Mb Magic Key 3 and EZFA 256Mbit
#49235 - headspin - Wed Jul 27, 2005 6:10 pm
| Chetic wrote: |
| Aren't you going to add a keyboard to it so you can write your own stuff though? |
It's what I want to do eventually, yes. The only tile based keyboard I've seen that I like is the DSLinux one by Pepsiman. I think I'll make one along these lines. I just hate deving on the DS at the moment because I have to test constantly on hardware, it's tiring sending over thu wifi, testing, sending over and so on.
_________________
Warhawk DS | Manic Miner: The Lost Levels | The Detective Game
#53346 - headspin - Wed Sep 07, 2005 7:33 pm
#53351 - zubiac - Wed Sep 07, 2005 7:53 pm
what?
So I can basicly type in what i want and the DS speaks my written words?
If true,I love you.
*calls Pizza-delivery service*(hehe)
EDIT: Works great
EDIT 2:
didn't work. She didn't understood salami."Pizza" worked though
Very nice app.
thanx a lot
_________________
Abusing Cube and DS with all sorts of homebrew and hacks.
#53353 - Dannon - Wed Sep 07, 2005 8:04 pm
Not too bad, needs some optimisation, but you know that already. Delete doesn't seem to work either because I keep putting in wrong characters and I want to get rid of them, should it work? Also, if I touch the screen below the keyboard I get loads of random characters appearing in the top screen, I don't know if you know about this, but now you do.
Keep up the good work
#53355 - headspin - Wed Sep 07, 2005 8:13 pm
Err, you caught me.. it's quite incomplete I know. Optimizations are definately needed along with a fix for the "off the keyboard" bug you mentioned (I completely missed it).
And your also correct, backspace dosn't work yet either :(
Ording a pizza would be difficult I would imagine, I can't write very quickly on that keyboard lol
_________________
Warhawk DS | Manic Miner: The Lost Levels | The Detective Game
#53358 - Dannon - Wed Sep 07, 2005 8:35 pm
If you could record the conversation that you're DS had with the pizza shop that would be cool, I'm sure it would make interesting listening for all
#53386 - chishm - Thu Sep 08, 2005 2:23 am
I have tried this and DS goo out, and for some reason the ARM7 is not working with my GBAMP NDS v2 loader. How are you communicating between the ARM7 and ARM9. Eg where is the IPC located, do you use the IPC fifo, etc. This will help in my debugging efforts.
#53408 - headspin - Thu Sep 08, 2005 7:00 am
Unfortunately, my demo's are based on dovoto's ndslib, not the CVS releases of libnds.
Check out www.drunkencoders.com/tools/ds/ndslib.zip
I think mine were based on ndslib\examples\arm7\soundPlay, ndslib\examples\templateBoth, or ndslib\examples\arm9\2D\Double_Buffer. But I'm not exactly sure which ones.
If you can't get them to work, I will finally install the latest DevKitPro + libnds and modify/rebuild them.
_________________
Warhawk DS | Manic Miner: The Lost Levels | The Detective Game
#53411 - chishm - Thu Sep 08, 2005 7:18 am
Amazingly all those demos work. However for some reason in your demos either the ARM7 is not starting properly or it is not receiving commands from the ARM9. I'd rather find out why they're not working than have you rebuild them, as I am aiming for 100% compatibility.
#53446 - brian33x51 - Thu Sep 08, 2005 2:25 pm
| tepples wrote: |
| headspin wrote: | you must understand permission must be granted by the original authors, even though Stefan practically re-wrote the synth engine. I do not know if he is still even in contact with them.
[...]
The source was not simply handed over to me by Stefan. I gave Stefan detailed correspondence regarding my intentions for the port. The problem here is commercial use of the code, which is against Stefan's wishes (I guess). |
Point him to the FFTW project, which has released a commercial-quality Fast Fourier Transform library under the GNU General Public License. The copyright owner agreed to do this only because the copyleft aspect of the GPL allows the copyright owner to price-discriminate against those who rely on a typical proprietary software business model. The LZO data compression library, used by PocketNES for its save states and by early versions of Tetanus On Drugs for GBA for its assets, is under a similar license: GPL to those who haven't paid and proprietary to those who have. |
I would avoid fftw for embedded.
Yeah, it's fast but it's go lots of bloat, in order to squeeze any and all performance out there's tons and tons of code.
Look into something like kissme FFT which is a much simpler (and therefore thrifty) implementation. His performance isn't so far below FFTW and his license is less restrictive if that matters.
#53486 - headspin - Thu Sep 08, 2005 6:39 pm
| chishm wrote: |
| Amazingly all those demos work. However for some reason in your demos either the ARM7 is not starting properly or it is not receiving commands from the ARM9. I'd rather find out why they're not working than have you rebuild them, as I am aiming for 100% compatibility. |
Thanks chishm, I have now uploaded new versions of DSGoo & DSSpeech built with the new DevKitARM so they should both work now with your GBAMP loader.
_________________
Warhawk DS | Manic Miner: The Lost Levels | The Detective Game
#53536 - chishm - Fri Sep 09, 2005 1:40 am
| headspin wrote: |
| Thanks chishm, I have now uploaded new versions of DSGoo & DSSpeech built with the new DevKitARM so they should both work now with your GBAMP loader. |
I can confirm that they work with the loader.
#53545 - cybereality - Fri Sep 09, 2005 2:17 am
headspin: I just checked out the DSspeech demo and its pretty cool. Reminds me of the old Speak & Spell I had as a kid. If youre into that suff, the AT&T Natural Voices are awesome. Check out Audrey (the UK Female voice). It almost sounds like a real person. It would be cool if these TTS engines could be used in a game to produce on-the-fly speech for say an RPG title. If the game characters were robots, for instance, it could make sense and add a cool feature to the game. Oh, and the DSGoo was pretty cool too, I just wish it worked in real-time.
_________________
// cybereality
#53570 - headspin - Fri Sep 09, 2005 7:55 am
If you like Speak & Spell, check out the Speak & Spell Flash Demo for a bit of retro fun.
I am definately into speech synths of all types and quite like AT&T Natural Voices. Like you said, they are getting really good at sounding real now.
You should hear the synthesized woman on our Telstra phone service here in Australia, it's really good and the voice recognition system is damn accurate (finally). I'm not sure how it handles accents or speech impediments though.
DSGoo does work in realtime if you change the pen size to small. But I havn't got around to optimizing it yet, so perhaps I can release a faster version in the near future.
_________________
Warhawk DS | Manic Miner: The Lost Levels | The Detective Game
#53742 - headspin - Sun Sep 11, 2005 11:37 am
- Backspace now working
- Pen off keyboard producing random characters bug fixed
_________________
Warhawk DS | Manic Miner: The Lost Levels | The Detective Game
#54319 - tssf - Sun Sep 18, 2005 7:34 am
I really like this thing..though I do have a couple suggestions.
The keyboard's buttons should dim, or darken when they're clicked, and there should be a click sound similar to the NDS firmware's keyboard.. I just think it would be better (it felt weird typing and not hearing anything)
Anyway this demo is really neat. It would be awesome to actually have a voice synthesizer kinda like Microsoft's Agent's, but that's probably impossible. Perhaps in the future if you do add different engines though, you can give the ability to choose between voices (or engines i guess would be a better way to put it)
Anyway, awesome demo :) I love showing these things off to my friends and have them go like, "Oh my god..that kicks ass, now I want a DS"
#54329 - headspin - Sun Sep 18, 2005 11:08 am
Nice ideas about adding the sound and highlight of the keys. I will add that when I get some spare time.
Shouldn't be too hard adding either, since the sound is just playing a sample, and the highlight is just switching palettes for the appropriate tiles.
_________________
Warhawk DS | Manic Miner: The Lost Levels | The Detective Game