G2

NK

Nagendra K.

Senior Engineer - Data Scientist

Enterprise (> 1000 emp.)

6/22/2021

"Speaker Verification using Kaldi Toolkit"

4.5/5

What do you like best about Kaldi ASR?

It is open-sourced and very well-maintained toolkit by the core group of Johns Hopkins University's speech recognition laboratory. We can extract various state-of-art features such as i-vector, x-vector which can be used for various speech-related tasks. For speech-related tasks, we can achieve the state-of-art result. Review collected by and hosted on G2.com.

What do you dislike about Kaldi ASR?

Installation of Kaldi with GPU support is a nightmare for me. Review collected by and hosted on G2.com.

Recommendations to others considering Kaldi ASR:

Before installing, first, read the Kaldi group discussion related to installation steps. Review collected by and hosted on G2.com.

What problems is Kaldi ASR solving and how is that benefiting you?

Mostly I have used Kaldi for the Speech-based Speaker Verification task. Training the Total-Variability matrix and Deep Learning model is much faster than the Matlab-based training. Review collected by and hosted on G2.com.