VoxForge
Ken,
I want to help out with support for Hebrew, and make a pronounciation dictionary (which you explain is necessary to make accoustic models) for Hebrew. I was wondering if you could send me the list of prompts that you are using in the submission app to start off with. Also, will HDK work in Hebrew, since it uses a different alphabet? Any help will be greatly appreciated.
Ethan
Hi Ethan,
>list of prompts that you are using in the submission app to start off with.
See this link (this is from the source code for the speech submission app).
>Also, will HDK work in Hebrew, since it uses a different alphabet?
I think you mean "HTK". If so, then yes it should work with Hebrew. HTK acoustic model creation is language agnostic, and my understanding is that it can work with any language.
The only thing that might be a bit tricky is that Hebrew is written from right-to-left, and I am not sure how that might impact things. Hebrew grammars might have to be written left-to-right for HTK (or Julius - since it uses HTK acoustic models) to understand them... There might be a switch somewhere in HTK. There might be something on this in the HTK email archives:
(note: you need to be registered with HTK to access these)
Ken
Hi Ethan,
On a side not, for some reason quite a few of your submissions did not include any audio files:
DE-ethanoconnors-19700101-bwp.zip - no audio
DE-ethanoconnors-19700101-rfj.zip - no audio
DE-ethanoconnors-19700101-vjp.zip - no audio
DE-ethanoconnors-20080605-agt.zip
DE-ethanoconnors-20080605-elz.zip
DE-ethanoconnors-20080605-eub.zip
DE-ethanoconnors-20080606-juy.zip - no audio
DE-ethanoconnors-20080606-nje.zip - no audio
DE-ethanoconnors-20080606-qft.zip - no audio
HB-ethanoconnors-19700101-khf.zip - no audio
HB-ethanoconnors-19700101-lli.zip - no audio
HB-ethanoconnors-19700101-xbw.zip - no audio
HB-ethanoconnors-20080605-bjh.zip
HB-ethanoconnors-20080605-qno.zip
HB-ethanoconnors-20080605-set.zip
HB-ethanoconnors-20080606-adb.zip
HB-ethanoconnors-20080606-qdi.zip
HB-ethanoconnors-20080606-xjv.zip
ethanoconnors-20080605-mpw.zip
ethanoconnors-20080605-oiz.zip
ethanoconnors-20080605-waa.zip
ethanoconnors-20080606-dqj.zip - no audio
ethanoconnors-20080606-grg.zip - no audio
ethanoconnors-20080606-nfp.zip - no audio
ethanoconnors-20080606-qzc.zip - no audio
ethanoconnors-20080606-rzz.zip - no audio
ethanoconnors-20080606-xnf.zip - no audio
I am not sure why this is occuring... The text information gets uploaded no problem (i.e. readme, prompts, license files), but there are no audio files.
Are you waiting for your submission to fully upload before starting to record another set of prompts? Are you using more than one browser window or tab to submit a recording? Are there any strange messages appearing? What Operating System, browser, Java run-time-environment are you using (please include version numbers)?
I need as much information as I can get so that I can try to replicate this problem, and then fix it.
thanks,
Ken