Since mobile hardware donn't have a fast cpu, normal speech recognition and synthesis
algorithms and engines may not suitable. It would call for a significant amount of processing power, not to mention storage capacity for the large databases.
There are of course other creative solutions that would need to be considered in implementing this type of system, such as server-based storage and processing or portable unit in the form of an external card.
Such applications would be particularly useful in communities that are traditionally oral, visually-impaired, illiterate and identity verification.
It looks like the obvious choice will be to implement HMM given its ease of use and implementation. HMM is the most reliable with respect to getting a usable result. You could start with Festival[1] and Flite[2] or with Sphinx[3] and PocketSphinx[4]. All packages Festival/Flite and Sphinx/PocketSphinx have been developed by CMU.
[1]http://festvox.org/index.html
[2]http://www.speech.cs.cmu.edu/flite/
[3]http://cmusphinx.sourceforge.net/html/cmusphinx.php
[4]http://www.speech.cs.cmu.edu/pocketsphinx/
No comments:
Post a Comment