exporting final model ready for serving after successful training

gooofy / zamia-speech

Open tools and data for cloudless automatic speech recognition

GNU Lesser General Public License v3.0

444 stars 84 forks source link

exporting final model ready for serving after successful training #94

Closed cogmeta closed 4 years ago

cogmeta commented 4 years ago

After succesfully finishing training with run-chain.sh, how to extract the tdnn_f model ready for serving with kaldi_decode_wav.py. Specifying the tdnn_f directory path in kaldi_decode_wav.py does not work.

gooofy commented 4 years ago

check out

speech_dist.sh

which can be used to generate distribution tarballs

cogmeta commented 4 years ago

perfect! it works.