Looks good Do you plan to let users choose models for the transcription?