Hy Cerin,

I can not answer your question, but I can tell you that if you need an acoustic model for Julius, I can tell you I get.

I'm working on a tool that uses speech recognition engine Julius.. I can not provide you with the software but I can give you the elaborate (the four files that are used to Julius to work:. dfa,. dict hmm15/hmmdefs and tiedlist)!

If you send me through mail ([email protected]):

-a file containing all the words that exist in the model (a .txt file in UTF-8 encoding);

-a .txt file inside with the corresponding text of each sound sample (prompts.txt)

-a folder containing the files you have recorded wave

I can build your own acoustic model used by Julius.

Bye C.

Thanks, that's kind of you, but I know how to make my own model. What I can't make is a huge corpus of speech samples from thousands of people :)

For data analysis, two subjective tests were held in subjective testing laboratory based in Prague, Czech Republic. They are named as A and B. Test A was performed in July 2015 and test B in January 2017.

>What's the status of the speech sample processing?

yes, I am behind in "processing" speech submissions, but you can still submit speech. 

>I wouldn't mind submitting more samples,

That would be great, thanks

>but if the Voxforge project has been abandoned (e.g. no news or updates since 2011)

The forums are active...

>or the samples are being lost, deleted, or unprocessed, I don't want to waste my time.

You can still make submissions, and they are stored on the VoxForge repository server (they are mirrored daily).  If you need access to the unprocessed submissions, let me know...

If you want to created an acoustic model using your own voice, and still want to help with open source speech recognition, you can use an audio recorder (like Audacity), record some prompts, package it up, and upload it to VoxForge using the old submission process.

In the future, it is planned to continue the investigation, experimenting with less complex parallel task within P.835 context or using different methodology (e.g., ITU-T P.800) for the existing parallel task. Also, standardization effort will be initiated to define parallel task subjective testing as a logical counterpart to traditional laboratory subjective speech quality tests.