An audio generation system receives audio instructions that have instruction channel designations and dynamically allocates synthesizer channels in groups of sixteen channels that support the MIDI standard to receive the audio instructions. The synthesizer channels are assigned to receive the audio instructions such that audio instructions having the same instruction channel designations are assigned to be received by synthesizer channels in different synthesizer channel groups. The audio instructions are routed to the synthesizer channels in accordance with the instruction channel designations of the audio instructions and the synthesizer channel assignments via mapping channels in a mapping component, where an individual mapping channel corresponds to a particular synthesizer channel.