Speech to Text on Kaldi
- 0 Collaborators
Intend to make kaldi run on AI dev cloud to make use of its multiple nodes and train in a distributed fashion. Leveraging Pytorch also for implmenting BiLSTM system as well. Need help in configuring system either using SLRUM/Grid engine/Some other method to run it beyond one node . Currently it might be able to run on one node of 24 cores but should be able to scale it to leverage other nodes as well. ...learn more
Project status: Under Development
Intel Technologies
AI DevCloud / Xeon,
Intel Opt ML/DL Framework
Overview / Usage
Kaldi and Pytorch can be used to build robust DNN based system for training your own speech to text system. We intent to work on it and make the system usable on AI dev cloud so that we could train in a distributed fashion
Methodology / Approach
We are looking on combine the existing open data sets and build on mutliple networks like RNN CTC, CLDNN and the new wave2letter architecture from facebook
Technologies Used
Kaldi
Pytorch
Arrayfire
Wave2letter
cereal
C++
python
bash