Back

2018-03-07 - Upgrading my IRC bot.

Recently, I've been extremely interested in Language processing, speech recognition, and other data processing.
I wrote a bot for an IRC server called Arcadia, you can check it out if you want: http://christianlifefm.com/chatrules.php


It has been quite a bit of fun modifying that script and improving it over time. One of the major changes, was splitting off the IRC part into its own thread.
It isn't a very complex bot, but it served its purpose well.
That being said, I have been wanting to upgrade it for some time.  I have been looking at ways, and toying around with ideas. IF I remember, I am hoping to write down what I've learned as I go.

There are a few things I've started looking into:
Tensorflow (Currently way too complicated for me to delve into, and not particularly my focus)
CMUsphinx - pocketsphinx
spaCy
Spark-NLP

So far, I have managed to get pocketsphinx working on windows, though it fails on FreeBSD, I'm not quite sure why yet. It won't run on my phone, because I don't have a new enough version of python.
I was able to get CMUsphinx to work on my phone, when compiled in Java using Android studio. There is example code available that makes it work pretty easily.
I've been playing around with that, and adjusting things. It matches key words really well, but fails on general language, so it won't work for dictation. I added the functionality of the phone reading back the text responses outloud via TextToSpeech

End goals:
- get Arcadia to respond better to text messages, maybe even talk back.
- Get voice translated to text, sent to Arcadia, and then sent back.

At the moment, I'm trying to get spaCy working. We shall see how that goes. :]

Edit post.