HARK for Speech Recoginition

Question

I'm at the research stage in designing a Speech Recoginition app using HARK open source (compel using HARK), but the HARK system documentation confuses me.

I have read the following HARK system documents:

• http://winnie.kuis.kyoto-u.ac.jp/HARK/document/1.2.0/hark-cookbook-en/index.html

• http://www.hark.jp/document/hark-document-en/index.html

I need a Speech Recoginition solution, and the HARK system appears to be an ideal complete app system for my robot audition project, because it includes many features, such as; Sound localization, Function transfer and Speech recoginition.

My problem is that I still dont know how the HARK system is intended to be used or how to use/communicate with the HARK open source.

I'm appealling to somebody with HARK system experience to help me get on the right track.

Answer 1

Hark works as a standalone server which sends features to the speech recognizer. Speech recognizer must accept features over TCP socket and process them. The relevant parts of documentation are:

SpeechRecognitionClient

JuliusMFT

You can also modify Hark sources and use it simply as a software library invoking functions and processing results.