Update

[EtienneAb3d]

What's new:

• input/output languages may now be different (to get a translation).
• some prompt examples that can be automatically used without the need to redefine new ones.
• model size can be specified.
• set isMusic=True to process music files with minimal filtering and no markers (not really interesting for music inputs)