It already exists. Modern clients like Mudlet (https://mudlet.org/) support both old (MSP) and modern audio protocols: https://wiki.mudlet.org/w/Standards:MUD_Client_Media_Protoco...
There's games like StickMUD that make use of this, too.
There's games like StickMUD that make use of this, too.
for a real audio based mud, the players voice input needs to be interpreted as commands but also sent to the other players in the room so you can have real voice based interaction.