Hi everyone! Happy new year!
I have a few questions about the project for 2019.
First I’d like to point out that I really liked the new UI, it’s easier now to record and validate, also stats are a really important part, because we can keep track of the progress and keep motivated, and I think with the ranking info people are more likely to contribute, It would be nice to see your own contributor position along side your contribution stats, I couldn’t find mine in the Top Contributors.
With regards to DeepSpeech: It seems to me that even though the datasets are publicly available, for most of us it’s not possible to train a DeepSpeech model, so we will depend on the official repo, I don’t know what are the specifics of the project, like, I assume the the model is a seq2seq, but the text normalization technique is it with words or chars? For example most likely people will use DeepSpeech on IoT with have little RAM, so using word2vec and a language model is not a option, that why I’m looking forward to see something similar to wav2letter on the DeepSpeech project and the release of models in tf lite format.