Voice recognition without Internet in Android

Question

I searched for several voice recognition libraries that support Android, but none of them work without a server call or without using the Internet. My requirement is to detect certain keywords and perform some actions accordingly without the Internet, i.e. Keyword analysis should be done locally. So, is there a voice recognition library that can do this. CMU Sphinx is the only toolkit that people say can accomplish this, but it still needs an algorithm to determine a keyword in speech.

Answer 1

You can use Pocketsphinx with either a statistical model that you can train on a body that reflects the language that users speak with your application, or you can define a JSGF grammar that explicitly defines the language as a set of context-free rules. In both cases, you need to have an acoustic model and a mapping of phonetic characters into spelling words. The CMU Sphinx project provides these resources for English and several other languages.

Once speech is recognized, it becomes a simple string to which you can apply any "spot keyword" algorithm that you want to apply.

An example of a Pocketsphinx-based Android application using JSGF-based speech recognition is Inimesed .