This model is trained on LibriSpeech dataset and can only be used for English ASR.

It's a very small model, which means it is suitable for embedded devices.