Click here to register.

English Speech Files

Nested
royerfa-20080307
User: royerfa
Date: 3/7/2008 4:26 am
Views: 1903
Rating: 17

Hello,

 

So If I understand, the best way to improve Julian speech recognition is to submit the maximum trainingvoice using voice.

And I update my julius-Quikstart with the new package of the day and Gone on Baby ...

 

Is that all we can do ???

And of course playing with the parameters in juilian.jconf

Regards,

good travel to the admin 

--- (Edited on 3/7/2008 4:26 am [GMT-0600] by royerfa) ---

--- (Edited on 3/9/2008 11:44 pm [GMT-0400] by royerfa) ---

royerfa-20080307.tgz royerfa-20080307.tgz

Notice: many prompts in "English Speech Files" were adapted from the prompt files contained in the CMU_ARCTIC speech synthesis database, which were in turn derived from out-of-copyright texts from Project Gutenberg, by the FestVox project at the Language Technologies Institute at Carnegie Mellon University.

Re: royerfa-20080307.tgz
User: kmaclean
Date: 3/9/2008 10:27 pm
Views: 185
Rating: 13

Hi royerfa,

>So If I understand, the best way to improve Julian speech recognition is to

>submit the maximum trainingvoice using voice.

If I understand what you are asking ...

Yes, the best way to improve Julian recognition for your speech is for you to add more speech samples. 

If you are looking for a non-GPL solution (something you don't plan to distribute) to improve recognition of your voice, you might look at using Keith Vertanen's HTK acoustic models and adapting these using your speech and VoxForge speech. 

>And I update my julius-Quikstart with the new package of the day and Gone on Baby ...

Yes, as your speech gets added to the VoxForge acoustic model in the nightly build, then Julian should get better at recognizing your speech.

>And of course playing with the parameters in juilian.jconf

That can make a very big difference in recognition performance too!

Ken 

 

--- (Edited on 3/9/2008 11:27 pm [GMT-0400] by kmaclean) ---


Notice: many prompts in "English Speech Files" were adapted from the prompt files contained in the CMU_ARCTIC speech synthesis database, which were in turn derived from out-of-copyright texts from Project Gutenberg, by the FestVox project at the Language Technologies Institute at Carnegie Mellon University.

Re: royerfa-20080307
User: kmaclean
Date: 3/10/2008 4:44 pm
Views: 373
Rating: 19

Hi royerfa,

Thanks for your submission!

See below for the particulars of your submission and a link to the audio in the VoxForge Speech Corpus. 

Ken

Speaker Characteristics:

Gender: [male];
Age range: [adult];
Pronunciation dialect: [French accent].

Recording Information (don't worry if you can't find some of this information):

Microphone make: [Andrea anti-noise NC-61];
Microphone type: [headset mic];
Audio card make: [Intel 82801CA-ICH3 with CS4205 at irq 5
];
Audio card type: [integrated]
Audio Recording Software: [Audacity rel 1.3.3 beta];
O/S: [Ubuntu].

File Info:

File type: [wav];
Sampling rate: [48kHz];
Sample rate format: [16bit];
Number of channels: [1];
Audio Processing: [n]

Prompts

vf7-1 I think it's much nicer to quarrel
vf7-2 I saw it when she rolled
vf7-3 I only read the quotations
vf7-4 He was the soul of devotion to his employers
vf7-5 Out of his eighteen hundred, he laid aside sixteen hundred each year
vf7-6 You have heard always how he was the lover of the Princess Naomi
vf7-7 They ought to pass here some time today
vf7-8 I had been sad too long already
vf7-9 All eyes, however, were staring at him in certitude of expectancy
vf7-10 He had observed the business life of Hawaii and developed a vaulting ambition
vf7-11 I may manage to freight a cargo back as well
vf7-12 O'Brien had been a clean living young man with ideals
vf7-13 He it was that lived to found the family of the Patino
vf7-14 Straight out they swam, their heads growing smaller and smaller
vf7-15 You won't die of malnutrition, be sure of that
vf7-16 See the length of the body and that elongated neck
vf7-17 They are coming ashore, whoever they are
vf7-18 Soaked in seawater they offset the heat rays
vf7-19 Think of investing in such an adventure
vf7-20 Nobody knew his history, they of the Junta least of all
vf7-21 I have been doubly baptized
vf7-22 They wouldn't be sweeping a big vessel like the Martha
vf7-23 Joan looked triumphantly at Sheldon, who bowed
vf7-24 And I hope you've got plenty of chain out, Captain Young
vf7-25 The discovery seemed to have been made on the spur of the moment
vf7-26 They handled two men already, both grub-thieves
vf7-27 Eli Harding asked, as Shunk started to follow
vf7-28 Now go ahead and tell me in a straightforward way what has happened
vf7-29 That's where they cut off the Scottish Chiefs and killed all hands
vf7-30 And after the bath a shave would not be bad
vf7-31 Now please give a plain statement of what occurred
vf7-32 You can take a vacation on pay
vf7-33 They are big trees and require plenty of room
vf7-34 And Raoul listened again to the tale of the house
vf7-35 There are no kiddies and half grown youths among them
vf7-36 Oolong Atoll was one hundred and forty miles in circumference
vf7-37 McCoy found a stifling, poisonous atmosphere in the pent cabin
vf7-38 It would give me nervous prostration
vf7-39 She said with chattering teeth
vf7-40 I'll be out of my head in fifteen minutes 

License 

Copyright (C) 2008  royerfa

These files are free software; you can redistribute them and/or
modify them under the terms of the GNU General Public License
as published by the Free Software Foundation; either version 3
of the License, or (at your option) any later version.

These files are distributed in the hope that they will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
GNU General Public License for more details.

[   ] royerfa-20080307.tgz 10-Mar-2008 12:53 25.2M

 

 

--- (Edited on 3/10/2008 5:44 pm [GMT-0400] by kmaclean) ---


Notice: many prompts in "English Speech Files" were adapted from the prompt files contained in the CMU_ARCTIC speech synthesis database, which were in turn derived from out-of-copyright texts from Project Gutenberg, by the FestVox project at the Language Technologies Institute at Carnegie Mellon University.

PreviousNext