Similar to this question. Since I don't want to pass the voice data to a server, doing this may cost me more time. I wonder if I can use HTK to recognize the voice data locally with the Android application so that I won't need to pass that audio to the server. Here may be the solution, but can anyone give me a more detialed tutorial on how to build HTK with android ndk? Thank you!
Asked
Active
Viewed 915 times
1 Answers
3
Maybe pocketsphinx
is more suitable for you, see this tutorial: http://cmusphinx.sourceforge.net/wiki/tutorialandroid . And to build the JNI API from pocketsphinx
yourself, see Building PocketSphinxAndroidDemo (from CMUSphinx project) for more details.

Bill Xia
- 299
- 2
- 14
-
Thank you for your help. Actually, we can compile HTK using NDK.:) – user978112 Oct 15 '14 at 10:13
-
Yeah, pocketsphinx is also not written in Java and should compile using NDK. I recommended pocketsphinx because it's more exquisite compared with HTK, and the accuracy is not bad. What's more pocketsphinx is more suitable for mobile devices. – Bill Xia Oct 15 '14 at 13:53
-
this does not answer the original question – Mehdi Souregi Oct 15 '19 at 10:37