Is it possible to have Expression-Based Dynamic Voice, as described here, with different models, e.g. 'tts_models/multilingual/multi-dataset/bark'?
'tts_models/multilingual/multi-dataset/bark'