Training Deepspeech throws missing characters

Please read other’s reports, there is already a ton of infos. What you see might not be what you have, utf8 data can be tricky, make sure you generate alphabet from your dataset