As currently defined, there is a two bit field to distinguish between the type of payload. There is an inference that voice-only means one particular type/rate of codec, and data+voice means another.
Codec2 is still an active project and it seems to me that M17 would be best served by adding another bit field to the type bits which indicates the particular type of codec. At the very least it would stop old firmware versions trying to decode newer codecs which they don’t understand. Typically newer firmware would include the old codecs already.
This question was raised on the MMDVM forum where someone suggested a HF gateway, and of course HF codecs are different, but in theory understandable by other Codec2 based systems without transcoding.