Index
> DOS > Talking program? |
| Author |
|
|
Artlav 28 Sep 2008, 17:43
I just dug up a gem out of an old floppy - a DOS program that speaks the thing given as an argument thru PC speaker, provided below.
Naturally enough, i got curious about how the thing works, yet the disassembler i have - an old IDA - broken it's teeth on it. Maybe someone know something about the program's origin, and how it could work? Or perhaps, got a working source code of a similar thing? Are there any modern non-MAC speech generators? Not like the half-life one, but a letter to sound one, like this.
|
|||||||||||
28 Sep 2008, 17:43 |
|
|
edfed 28 Sep 2008, 19:26
for XP users.
in config board: sound/ voice/ "you have selected sam as the microsoft default voice" just test, it have a better quality than SPP.exe. i don't know what it is exactlly, but is on all XP installation. |
|||
|
28 Sep 2008, 19:26 |
|
|
Grom PE 29 Sep 2008, 06:23
Yes, I know this little program too.
It's amazing that it can talk yet is just 44 kb in size. Works only in pure DOS or DOSBox because of complex PC Speaker usage. It speaks words from command line (in Russian CP866, but you can try English too), uses + after vowel for accent (behaves strangely if it's not a vowel), understands ?, !, : in the end of sentences and changes pitch accordingly. It makes too long pause on space, so I use - or _ to split words visually. I wanted to understand how it works, too. Having sources would be awesome. Try changing byte in spp.exe at 0000A0A2h from 77h to 60h, it will talk faster =) Use following .bat file to see how it works: Code: @echo off spp wse+m-priwe+t spp hello+-me+n!-ha+urju+du+i+n? spp spi+kinzi+do+iv? da+si+st-fanta+stiv |
|||
|
29 Sep 2008, 06:23 |
|
|
DOS386 30 Sep 2008, 06:05
Artlav wrote: DOS program that speaks the thing given as an argument thru PC speaker, provided below. YES it works (since I am the only one here in who has DOS at all ...). But the quality is very bad ... could be much better, keeping both DOS and PC speaker of course. 
Quote: Naturally enough, i got curious about how the thing works, yet the disassembler i have - an old IDA - broken it's teeth on it. Maybe someone know something about the program's origin, and how it could work? Very trivial thing ... just some work with letter mixing and some samples of a nice voice
> It's amazing that it can talk yet is just 44 kb in size. It's amazing to see how the bloat of some fine software products is racing towards the infinity  _________________ Bug Nr.: 12345 Title: Hello World program compiles to 100 KB !!! Status: Closed: NOT a Bug |
|||
|
30 Sep 2008, 06:05 |
|
|
Borsuc 30 Sep 2008, 11:38
DOS386 wrote: It's amazing to see how the bloat of some fine software products is racing towards the infinity 
even though of course, the actual "samples" (if you want high quality) will make most of the size. The bloat will still be there, and I hate bloated ****, but the size will be bigger even without bloat. unless you use compression -- FLAC for example is completely lossless _________________ Previously known as The_Grey_Beast |
|||
|
30 Sep 2008, 11:38 |
|
|
DOS386 04 Oct 2008, 05:37
> actual "samples" (if you want high quality) will make most of the size.
Right. But the primary problem of the version above seems not bad quality of input samples, but rather an unprecise speaker control or too low output sampling frequency, and too high volume. > unless you use compression -- FLAC for example is completely lossless Right ... but bloated, Vorbis (lossy) decoder is also bloated, so WAVPACK might be best here |
|||
|
04 Oct 2008, 05:37 |
|
|
ManOfSteel 04 Oct 2008, 08:33
I tried it in full DOS mode, but it doesn't work. I only hear high-pitched fast noise like a modem initiating a connection. Why's that? As far as I know, the PC speaker works quite well for other applications.
Artlav wrote: Maybe someone know something about the program's origin, and how it could work? Or perhaps, got a working source code of a similar thing? Although it doesn't work here for some reason, I can guess it's using formant synthesis. Some two years ago, I tried making a speech synthesizer using formant but abandoned it. The information on how to apply the theory is quite scarce. But basically you have to simulate the entire vocal tract. This method gives quite bad and unnatural results but you can have very high control over the output (intonation, timbre, gender, age, ...) |
|||
|
04 Oct 2008, 08:33 |
|
|
DOS386 04 Oct 2008, 23:44
ManOfSteel wrote: tried it in full DOS mode What is "full DOS mode" ? What kernel ? Quote: But basically you have to simulate the entire vocal tract. This method gives quite bad and unnatural results but you can have very high control over the output (intonation, timbre, gender, age, ...) Right ... and no need for bloated samples or FLAC or WAVPACK  _________________ Bug Nr.: 12345 Title: Hello World program compiles to 100 KB !!! Status: Closed: NOT a Bug |
|||
|
04 Oct 2008, 23:44 |
|
|
ManOfSteel 05 Oct 2008, 07:35
Quote: What is "full DOS mode" ? What kernel ? An MS-DOS shell not "emulated" by Windows, i.e. choosing the 'Command-Prompt' option from the startup menu ( F8 ) in a Win9x system (98SE). |
|||
|
05 Oct 2008, 07:35 |
|
|
tikbalang 01 Dec 2008, 03:32
here are three more similar dos programs. if anyone has the original archives with docs please let me know.
|
|||||||||||||||||||||||||||||||
|
01 Dec 2008, 03:32 |
|
|
DOS386 03 Dec 2008, 03:28
tikbalang (welcome to the FASM forum
) wrote:
> here are three more similar dos programs. TRAN is corrupt 
MLUV lacks TFM and doesn't work (too fast CPU ?) (C) 1990 CZ ? ATOM ... hehe 
> if anyone has the original archives with docs please let me know Indeed ... I don't |
|||
|
03 Dec 2008, 03:28 |
|
|
tikbalang 03 Dec 2008, 11:09
- attached 2 un-compressed versions of tran.
"say.txt" came from another text-to-speech dos program that worked better than what i already uploaded here. i just need to remember what it was and where i put my copy. for music, early MOD trackers can play thru the pc speaker (fasttracker2, modplay). modplay is still available here: http://www.awe.com/mark/dev/modplay.html there are even soundcard emulators like "remus" and "virtual sound blaster" than can output to the pc speaker, covox among other things. remus supossedly can emulate gus, sb, awe32, adlib, dac, dac stereo and output to existing card, pc speaker or covox. i'm unable to test it, however, since my onboard audio seems to interfere. i was able to test and use vsb succesfully back when soundcards were still expensive accesories. it can only emulate SB 1.xx (mono). the code was released free by its author Andrew Zabolotny. i can upload them here if anyone wants it.
|
|||||||||||||||||||||
|
03 Dec 2008, 11:09 |
|
|
baldr 03 Dec 2008, 13:57
tikbalang wrote: …than can output to the pc speaker, covox… |
|||
|
03 Dec 2008, 13:57 |
|
< Last Thread | Next Thread > |
Forum Rules:
|