No not quite. im talking more about software voices that can call upon any of the hardware channels (in gb case pu1, pu2, wave/user defined waveform and noise.) rather than only one.
Maybe it is similar if each of the 3 c64 channels can be a square, noise etc and are not hardwired channel 1: pulse, channel 2: tri, ch3:noise.
If that is the way it works, the difference is, each of those c64 voices can only trigger one of the sound channels at once, and not all of them together.
PR8 for the nes has voices that use up to four of the availible channels, in the format of a step sequencer.
you assign the channels you want to use for a particular voice, along with parameters such as the pitch, vol, pan etc. Ie;
So, in one step sequencer track (a single trigger in fact), it can trigger the hw channels; pulses, tri, dpcm and noise together, if they are all associated with that voice. (Contrast this with nanoloop, where any sequencer track can only access its associated gameboy sound channel (r,l,s,n being tied to pulse1, pulse 2, wav, noise)
Its quite possible to have triggers for two or more voices that use the same channels simultaneously, and in that case the voice furthest down or up the step seq gets priority for the channels that are trying to be accessed by more than one voice.
All i am imagining, is the same thing with the gameboy, except cutting out the step sequencer and controlling it via midi.
It would require a parameter that lets you set a priority level to the voices, one to set them a midi channel, one that lets you link any number of the hardware channels, and then whatever params such as root note/ pitch, volume, etc etc for each of the associated hardware channels in that voice.
Dya follow? )
*and making the wav channel sound exactly like pulse channels is totally unrequired! 
this is positively embarrassing.