Hi, I’m fairly familiar with audio processing and deep learning but not deeply experienced with speech/ASR/Kaldi. I might need to work on ASR for speech data analysis (e.g., topic modelling etc) as an application, not as research (it’s a company project). Should I Kaldi or not? I’m hesitating to say ‘yes’ to myself because my shallow impression is that it’d take me quite a while to learn how to use, but I’m not sure speech will be my very topic in the future.

​

FYI, if these are relevant, I don’t think we have lots of annotated dataset by ourselves, so I probably should rely on public datasets. At this moment there’s no demand for multiple languages.