The default model can be used to transcribe any audio type.
The command and search model is optimized for short audio clips, such as voice commands or voice searches.
The default and command and search recognition models support all available languages.
