From official documentation
Flexibility of sphinx4 allows you to build such a system quickly. It's easy to embed sphinx4 into flash server like red5 to provide web-based recognition, it's easy to manage many sphinx4 instances doing large-scale decoding on a cluster.
On the other side, if your system needs to be efficient and reasonably accurate, if you are running on embedded device or you are interested in using recognizer with some exotic language like Erlang, pocketsphinx is your choice. It's very hard to integrate Java with other languages not supported by JVM pocketsphinx is way better here.
in conclusion
TL;DR - use pocketshpinx
Use a C library if you plan to develop application on multiple platforms. Both Android (via NDK) and iOS(natively) support loading and calling C library. There is no JVM on iOS, thus, you won't be able to use sphinx on iOS.
To save time, pocketphinx-android-demo github repo has precompiled pocketsphinx libraries for all the relevant platforms - https://github.com/cmusphinx/pocketsphinx-android-demo/tree/master/app/src/main/jniLibs
You get both performance and potability(you have to compile for arm and x86) with C/C++ libraries. Here is a Dropbox, Inc. talk on how they use C++ for writing multi platform applications. https://www.youtube.com/watch?v=ZcBtF-JWJhM
Sphinx4 and PocketShpinx are completely different regarding implementation, you have to rewrite the demos or worse they might be not supported in PocketSphinx. That happens if you use some Sphinx4 advanced features.
If you wrote a JNI wrapper around pocketsphinx, then you could use that from your java code.
There are also python language bindings for pocketsphinx.
Essentially pocketsphinx is the speech recognizer you should use if you want a C language speech recognizer (and is also the one designed to be fast enough to work on a mobile phone).
