Do i need noise reduction algorithms while developing language model for my local language

Question

I am developing an acoustic model and an language model for my local language and I have planned to integrate it with pocketsphinx and sphinx4.

Now I want to ask do i need to implement an noise reduction algorithm for transmission accuracy for my local language or the pocketsphinx will take of that ? If yes I need to, then what should be the steps I should follow for integrating the noise reduction algorithm ?

Answer 1

You do not have to use any noice reduction algorithms.

If you do use one, make sure you apply it to both the training audio and recognized audio. Otherwise you may notice a drop in accuracy.

Depending on the algorithm, you may be losing some important data if you use it. The sound may seem better for you, as a human, but it may become harder to recognize for Sphinx.

Answer 2

Not necessary. Noise reduction is unrelated feature. If you do not have noise there is no need for noise reduction.