Run my own model

Hi!! I built a quechua language model 2 years ago, but I used deepspeech version: v0.2.0-alpha.8.0-gcd47560.

I run my asr using:

deepspeech output_graph.pb quz_alphabet.txt result.wav

but, now I am using Deepspeech version: 0.7.4 which it has another structure.

What I should do?

If you want to use 0.7.4 (or 0.8.0, which got released yesterday), you’ll have to retrain the model.

How can I install deepspeech version v0.2.0-alpha.8.0-gcd47560?

Just train on 0.8, there’s no point in using that old version.