How to extract MFCC features in PocketSphinx on Android

Question

I recently downloaded the PocketSphinx Android Demo for Android Studio. It worked on my Galaxy S5 and I'm actually surprised about the accuracy. However, I'm struggling to extract MFCC features for several reasons:

1. There is an explanation how to use the FrontEnd class to generate MFCC features but it is written for the Sphinx-4 implementation. How and where should I implement the sphinx property file which contains lines such as:

   <"component name="mfcFrontEnd" type="edu.cmu.sphinx.frontend.FrontEnd">

2. This leads to the question how I can use Sphinx-4 libraries in PocketSphinx?

Answer 1

There is an explanation how to use the FrontEnd class to generate MFCC features but it is written for the Sphinx-4 implementation. How and where should I implement the sphinx property file which contains lines such as: <"component name="mfcFrontEnd" type="edu.cmu.sphinx.frontend.FrontEnd">

If you are going to use sphinx4 jar in android, you can put xml configuration inside the jar as a resource. Or you can put it in assets.

This leads to the question how I can use Sphinx-4 libraries in PocketSphinx?

Pocketsphinx and sphinx4 are different implementations of the speech recognition engine, one can be replaced by another, there is no much sense to use them together, you need to select one one them.