As of now, is Deep Speech viable for real-world applications?

Yes, that is the approach I’ve used in the past to avoid cutting into the beginning of the utterance. It worked fairly well for my use case.