VoxForge
VoxForge
In Step 2, here gives the contents of the file prompts. In this tutorial we can just copy them and create the required file. But if we create a new project, how to create the prompts? Manually or by some HTK command?
Now, I create a new project . When I execute the command HDMan, and the file dlog shows that so many phones have ONLY Y one occurence. In the tutorial here gives the tips:"if there are phones that only have one occurence, you must add words that use these phones to your pronunciation dictionnary. You can search through the lexicon file for the phones you need, and then include the word that contains that phone. "
Does this means that we should add additional words to the prompts? Or add to the created file "dict"? I miss that.
In my opinion, if the "prompts" file contains so many words that not only the words appear in the grammar, then the Pronounciation Dictionary may reach to the requirement.
Hi Shawn,
>But if we create a new project, how to create the prompts? Manually or by some HTK command?
It depends.
If you created your own grammar file, and are only interested in creating acoustic models for this grammar file using your speech, then theoretically all you need is to use HTK's HSGen command (see Step 2 of the "Testing Your Acoustic Model with HTK & Julius" how-to) to generate prompts, which you then record using your voice. This works fine for creating monophone acoustic models. You are essentially creating an acoustic model containing hmms for just the list of words in your prompt file.
However, to create more robust triphone acoustic models, the VoxForge tutorial creates hmms based on all the words in your pronunciation dictionary - i.e. using words that are *not* just in the list of words in your prompt file. If you do not have enough words in your prompt file then HTK might not have enough speech to compile your acoustic model.
My recommendation, if you want to use prompts generated from your own grammar file, is to create a list of prompts using the HSGen command, but supplement you prompts list with the prompts shown in Step 2. This way, you get good speech coverage for your grammar, and you also have good coverage of the phonemes to at least compile your acoustic model.
>In the tutorial here gives the tips:"if there are phones that only have one occurence, you must add words that use these phones to your pronunciation dictionnary.
This should read: "you must add words that use these phones to your prompts file".
>Does this means that we should add additional words to the prompts? Or add to the created file "dict"? I miss that.
Add the words to your prompt file - they will get automatically added to your dict file.
>In my opinion, if the "prompts" file contains so many words that not only the words appear in the grammar, then the Pronunciation Dictionary may reach to the requirement.
I'm not sure I understand what you are asking here ...
Ken