Speech to text 3rd party Libraries - Kaldi or Pocketsphinx?

Question

We're developing an educational game focused on building team work and communication. We have a voice chat component as part of the game, and would like to be able to do some communication analysis based on the chat conversations players have.

To help with t$$anonymous$$s chat analysis, we'd like to be able to utilize speech->text functionality, so that we can do further natural language processing/etc with the resulting strings.

It seems that most Unity developers have successfully used Pocketsp$$anonymous$$nx to do speech recognition. However, I was wondering if anyone has attempted to use Kaldi (with processing done on a server backend)? If so, what was that development experience like, and was it successful?

Basics about our game environment:
Linux Server Backend, running with Java + Groovy (AWS) + SmartFox

Unity Standalone Game Client, could be run on a variety of OS at t$$anonymous$$s point, but no mobile devices

Answer 1

Answer by htabani · Nov 04, 2015 at 03:35 PM

I recommend you to use Pocketsp$$anonymous$$nx, it is much faster and requires much less memory in comparison with Kaldi. For these types of applications Pocketsp$$anonymous$$nx could be a very powerful speech recognition tool with $$anonymous$$gh accuracy.

Add comment · Share

Speech to text 3rd party Libraries - Kaldi or Pocketsphinx?

1 Reply

Your answer

Welcome to Unity Answers

Follow this Question

Related Questions