Ask HN: I want to get started with Speech-to-Text. Where do I begin?

5 points by thejoker20 4y ago ↗ HN

I am a novice when it comes to Machine Learning, but I think that having a subject to study, Speech-to-Text, will keep focused.

I recently started looking at CMUSphinx, Kaldi and DeepSpeech.

Do you have any tips for me?

2 comments

[ 21.5 ms ] story [ 156 ms ] thread

ginkoutest 4y ago ↗

Coqui (https://github.com/coqui-ai) is a great open-source STT resource you could start with. They have a lot of docs explaining how everything works and has a low barrier to entry.

ranuzz 4y ago ↗

As part of ETL or just basic understanding about the how the speech data is handled try this tool : https://github.com/wiseman/py-webrtcvad

It is a python wrapper for a library for voice activity detection. It acts as a starting point while working on speech recognition problems. Helped me understand and discover a lot of concepts related to audio signal and data when I was in your shoes.