Language Model for CMU07a.dic

[anam]

To use PocketsphinxController, we need a language model and a phonetic dictionary for it. If I want to use default CMU07a.dic, how can i generate language model for that?

[Halle Winkler]

Welcome,

You can’t use an 80,000 word language model; it won’t recognize accurately or in a reasonable timeframe and it will crash. For offline recognition on the iPhone you probably shouldn’t use more than 200 words in a single language model for best results.

[anam]

Thanks for your prompt reply. I made my own dict file but i’m getting this warning on console:

sih_add WARNING: repeated hashing of ‘THING’, older value will be overridden.
sih_add WARNING: repeated hashing of ‘WELL’, older value will be overridden.
sih_add WARNING: repeated hashing of ‘WOULD’, older value will be overridden.

I thought i put these words twice times in txt file but that is not the case. Can you please tell me what is that?

[Halle Winkler]

Why not use LanguageModelGenerator to create your language model?

[anam]

Yes, I’m using LanguageModelGenerator, giving txt file as input and getting its dic and languagemodel files. But I’m receiving “sih_add WARNING: repeated hashing” for many words.

[Halle Winkler]

OK, don’t worry about that warning, it isn’t critical.

[anam]

Thanks for your instant support.

[Author]

Posts