Natural Voice Dictionary for NS TTS

Take a script, leave a script - Like the penny tray only different

Natural Voice Dictionary for NS TTS

Postby cp99 » Fri Jun 02, 2006 4:57 pm

Many of us use the TTS script for NetStumbler with the default MS TTS engine. This can cause problems like mispronounced SSID names, turning SE into South East, and spelling out all capitol letters.
AT&T Natural voices, a wonderful TTS engine can be installed, and used in its place.
Demo here: http://www.research.att.com/~ttsweb/tts/demo.php

This TTS engine still has many pronunciation problems, which is why I have compiled a dictionary of common and default SSID, and their proper (or close) pronunciation to be used with Natural Voices.

You can here a demo of how it would sound while running in NS with the TTS script http://depositfiles.com/files/90118/ttsdemo.wav.html

Attached is a text file, to use simply, paste the text into en_us.dict (or your local language) in the NV directory.

If you use Natural Voices, then this dictionary is for you, and if you don't use Natural Voices, It is something to look into.

EDIT***

If there is an SSID missing, or one you want added to the list to be fixed for use with NV, please let me know.


CP
Attachments

[The extension txt has been deactivated and can no longer be displayed.]

User avatar
cp99
Mini Stumbler
 
Posts: 35
Joined: Sat Feb 21, 2004 8:15 pm

Postby goldfndr » Thu Jun 29, 2006 12:25 am

I have to say, this is a really nice list. It's a nice complement to my own list (attached). Mine is geared toward generating WAV files from the demo site rather than using a local installation of the licensed product.

Using your list, one can generate WAV files from the site. All it takes is changing { to <pron sym=" and changing } to " />. For example, GlobalSuiteWireless (which I found rather daunting) would be entered as
Code: Select all
<pron sym="g l ow 1 b ax l 0 s w iy t w ay 1 er l ih s 0" />
and sounds much better than the "Glow Bals White Wireless" that Microsoft Sam speaks.

I had a desire to learn the IPA alphabet as used by AT&T Natural Voices, and I can definitely learn from you. You'll see there are some places where we disagree on pronunciation (e.g. MSFTWLAN) but I don't know of any authoritative SSID pronunciation guide. (as if!)

Some other potentially recognized directives (according to the documentation) are:
  • <rate absspeed="n">text</rate>
    (n being 18 for 8x speed, 10 for 3x speed, 6 for 2x speed, 0 for 1x speed, -6 for 1/2 x speed, -10 for 1/3 x speed, -18 for 1/8 x speed, or integers between)
  • <silence msec="n" />
  • <spell>text</spell>
  • <voice required="name=voicename">text</voice>
    (voicename being one of the voices in the list, e.g. audrey)
  • <volume level="n">text</volume>
    (n being between 0 and 200 inclusive, with 100 for normal, 200 for twice as loud)
pron is for PRONunciation.

Hmm, need to get a "Safeway" recording for 5ECUR3w3p5TOR3 next... oh, and the next one I'd like to see is "orange" (and its siblings orange12/orange13/orange14).
Attachments

[The extension txt has been deactivated and can no longer be displayed.]

goldfndr
 
Posts: 181
Joined: Sat Aug 28, 2004 3:30 am
Location: Bellevue, WA, USA

Postby G8tK33per » Thu Jun 29, 2006 12:56 am

goldfndr wrote:pron is for PRONunciation.


Damn...
"Make yourselves sheep and the wolves will eat you." ~ Benjamin Franklin

Sons of Confederate Veterans
User avatar
G8tK33per
 
Posts: 6078
Joined: Fri May 09, 2003 4:00 am
Location: Goomba's Booty Boardwalk

Postby Dutch » Thu Jun 29, 2006 1:57 am

goldfndr wrote:I have to say, this is a really nice list. It's a nice complement to my own list (attached). Mine is geared toward generating WAV files from the demo site rather than using a local installation of the licensed product.

Using your list, one can generate WAV files from the site. All it takes is changing { to <pron sym=" and changing } to " />. For example, GlobalSuiteWireless (which I found rather daunting) would be entered as
Code: Select all
<pron sym="g l ow 1 b ax l 0 s w iy t w ay 1 er l ih s 0" />
and sounds much better than the "Glow Bals White Wireless" that Microsoft Sam speaks.

I had a desire to learn the IPA alphabet as used by AT&T Natural Voices, and I can definitely learn from you. You'll see there are some places where we disagree on pronunciation (e.g. MSFTWLAN) but I don't know of any authoritative SSID pronunciation guide. (as if!)

Some other potentially recognized directives (according to the documentation) are:
  • <rate absspeed="n">text</rate>
    (n being 18 for 8x speed, 10 for 3x speed, 6 for 2x speed, 0 for 1x speed, -6 for 1/2 x speed, -10 for 1/3 x speed, -18 for 1/8 x speed, or integers between)
  • <silence msec="n" />
  • <spell>text</spell>
  • <voice required="name=voicename">text</voice>
    (voicename being one of the voices in the list, e.g. audrey)
  • <volume level="n">text</volume>
    (n being between 0 and 200 inclusive, with 100 for normal, 200 for twice as loud)
pron is for PRONunciation.

Hmm, need to get a "Safeway" recording for 5ECUR3w3p5TOR3 next... oh, and the next one I'd like to see is "orange" (and its siblings orange12/orange13/orange14).

Take a look at Festival, and the runtime flite engines. Enclosed is a zipped wav file of flite output for GlobalSuiteWireless, using the default flite voice. You can get better output by recompiling flite with the higher quality voices, or using Festival.

Dutch
Attachments
gsw.zip
(49.29 KiB) Downloaded 391 times
All your answers are belong to Google. SEARCH DAMMIT!
Warning. Warning.
Low C8H10N4O2 level detected. Operator halted....
User avatar
Dutch
 
Posts: 6698
Joined: Fri Mar 05, 2004 12:00 pm
Location: City of Mermaids, Denmark


Return to Scripts

Who is online

Users browsing this forum: No registered users and 4 guests

cron