Hi, I’d like to know if there’s plans for a Windows bindings for Python or NodeJS, also I guess only the Mozilla team is able to train these models at the moment and I think the community would gain a lot with more easy-to-use libraries for example to get a model with an URL like we’ve seen with TF-Hub and PyTorch-Hub.
english_model = deepspeech.hub(“https://deepspeech.hub/model?=common_voice_english_v2.tflite”)
Thanks in advance!!
We already have Python bindings on Windows. And anyone can train a DeepSpeech model, we have a section explaining the process in the main README file.
Can you describe how much that does help ? Honestly, that seems like a huge amount of work for a feature that is not going to help a lot.
How are our current model URL’s, e.g.
https://github.com/mozilla/DeepSpeech/releases/download/v0.5.1/deepspeech-0.5.1-models.tar.gz
insufficient? Use of our current URL’s and the URL’s you suggest differ by only a few lines of code.
I think that’s enough, @reuben I know that anyone can train in theory, but I does require a lot of resources so a centralized model would be ideal, I’ll test now the pretrained model.
Thank you guys for responses. It was very helpful.
I still don’t understand the added value of what you pointed in the first topic, given that anyway you will need to use libdeepspeech somehow.