VoxForge
VoxForge
когда последий раз пробовал пополнить базу - у меня ничего не вышло
тоесть аплет загрузился
записи записал
а вот кнопке отправить ничего не отправлялось
аплет по всей видимости не мог соединиться с сервром для отправки данных...
собственно на мой взгляд есть проблема с доступом к этому сайту. Дропаются пакеты...
на voxforge посути можно выложить свой голос, но судя по всему нет возможности выложить порезанные на части аудио-книги (проблемы с правообладателями могут возникнуть).
я создал програму которая помогает нарезать аудио-книги и могу создать git репозитарий для размещения нарезанного материала.
Есть желающие присоединиться к этим начинаниям?
jabber/email: [email protected]
jabber/qip: [email protected]
Google translate:
actually I think there is a problem with access to this site. Start dropping packets ...
on
voxforge posuti can put your voice, but apparently there is no way to
put chopped to pieces of audio-books (the problems with copyright
holders may arise).
I created a program's which helps to cut the audio books and can create a git repository for placing chopped material.
There are wishing to join these endeavors?
Google translate is not clear on wheat your program does... please elaborate...
does you program segment audio books and put the results in a git repository?
thanks
Thank you for reply kmaclean
I'am creating project help cut/split audio-book in part (10 - 30 seconds)
and creating russian language model
project contain modules:
1) https://github.com/zamiron/ru4sphinx/tree/master/split
core spliter module, need perl, sox and sphinx3 (support any language in theory)
2) https://github.com/zamiron/ru4sphinx/tree/master/split/msu_ru_zero.mllt_cd_cont_2000
my last russian voice model for sphinx
TOTAL Words: 79878 Correct: 74804 Errors: 6076
TOTAL Percent correct = 93.65% Error = 7.61% Accuracy = 92.39%
TOTAL Insertions: 1002 Deletions: 1628 Substitutions: 3446
3) https://github.com/zamiron/ru4sphinx/tree/master/text2dict
russian transcriptor module, need perl
contain russian dictonary accent
it program creating .dic for sphinx
4) https://github.com/zamiron/ru4sphinx/tree/master/text2norm
russian text normalization
5) https://github.com/zamiron/ru4sphinx/blob/master/addons/linguistic_questions
russian linguistic_questions for sphinxtrain
>I'am creating project help cut/split audio-book in part (10 - 30 seconds)
>and creating russian language model
very cool, thanks for letting us know!
Are you using forced alignment with Sphinx for the segmentation of the audio book?
thanks,
Ken
1) for siliens detect, segmantation and convert wav/mp3 using "sox" yet
2) for spliting text using "sphinx3_decode" and "sphinx_jsgf2fsg"
logic aligment:
1) genetare JSGF grammar
example:
#JSGF V1.0;
grammar all;
public <all> = бедной из [ берёзовых ] [ кругляшей ] [ с ] [ бочками ] [ старой ] [ железной ] [ кроватью ]
2) convert jsgf to fsg and use for aligment with sphinx3_decode
3) change jsgf and try againg for best seach best result
Google translation:
when successive times, tried to fill up the base - I did not work
ie an applet loaded
records recorded
but the button to send anything not sent
applet apparently could not connect to servrom for sending data ...
I am hoping that this was once only problem - please try to submit again,
thanks,
Ken
i try again, but do not work :(
i using transparent proxy squid
last log:
1302938074.884 16333 10.X.X.X TCP_MISS/200 51145 GET http://www.voxforge.org/ru/listen - DIRECT/99.249.219.226 text/html
1302938077.273 643 10.X.X.X TCP_MISS/200 9724 GET http://www.voxforge.org/home/downloads/speech/russian/zamir-20100429-xmi? - DIRECT/99.249.219.226 text/html
# telnet 99.249.219.226 80
Trying 99.249.219.226...
Connected to CPE0080c813a40b-CM000f212fc2ef.cpe.net.cable.rogers.com (99.249.219.226).
Escape character is '^]'.
quit
<html><body>
<skip>
</body></html>
Connection closed by foreign host.
>i using transparent proxy squid
From the Java Applet Troubleshooting Guide: If your browser is behind a proxy server, please set up your Java JRE as follows:
Proxy Servers