Getting res="" when training common voice tamil data

[othiele (Olaf Thiele)]

I guess you are using the 0.6 branch.

KenLM ist used to generate the words.arpa and lm.binary and you use the Deepspeech native client to generate the trie file. And you will use the lm.binary and trie for training.

If -o 3 is returns an error, use 2. It is to find common word combinations and the more the better, but you want to get going

And don’t build the native client, download it:

Trie file creation

Please state what version you are using, as in the latest release you don’t longer need a trie file. But it looks like you are using 0.6x, so you will need to get the native client. Try python3 util/taskcluster.py --target native_client in the Deepspeech directory.