Conference napalm::commusic_v1

1968.0. "Dispense with Quantization (someday)" by ANT::JANZEN (T - 500 picoseconds and counting) Thu Apr 20 1989 20:42

	Wouldn't it be swell to get away from clock ticks and measures
    altogether in MIDI sequencers?
    
        Which would be better?  To have a clock tick be the ultimate measure
    bar  lining up all events in a piece, or to have real numbers (floating
    numbers, exponential notation numbers) in seconds define the time
    before the next event on a particular line (separate lines are independent).

    With the latter, it would be possible to notate pieces that have
    a free rhythm the way they are written.  It's one thing to say that
    you could make a score that approximated a piece by Conlong Nancarrow,
    in which separate lines have tempos in an irrational proportion,
    and another to say that the sequencer notation allowed you to set
    separate tempi in separate lines (staffs, perhaps), say in the ratio
    100 to 1.414...  Or close enough that audible rounding error did
    not appear in 2 minutes or 10 minutes.
    Or be able to notate Reich's Piano Phase the way it is written and
    get it played correctly? (two pianos playing the same phrase at
    different tempi, say for example MM 60 and MM 59.7 (I calculated
    a real example for the piece).  Isn't this better than calculating
    the quantizaton differences all the way through a 12-minute version
    of the piece?
    
    Let's hear it for floating numbers and separate tempi on separate
    lines.
    Two possible implementations might involve either the computer
    calculating the time to the next event on any staff, and using one
    time counter to time it out, or using separate counters to quantize
    separate lines, interrupting the processor for each next event.
    Tom

    Minor problem: 
                                        
    incremental time :== "play this note nn.nn microseconds after the
    			last note on this channel"
    
    absolute time :== "play this note nn.nn seconds after the start
    			of the piece.
    
    If you use "incremental time" then roundoff errors
    will tend to accumulate, especially between two tracks played
    simultaneously but without inter-track synchronizations.
    
    If you use "absolute time" then you can't "subroutine" choruses
    and fills.  Every musical event can happen only in one place in
    the piece.                       
    
    
    Conclusion:
    
    	an advanced sequencer must support the concept of a "SYNCH"
    	operation between two tracks. 

    Sounds like a SMPTE - oriented sequencer.  I haven't paid too much
    attention, not being in the video biz, but there are several programs
    like Opcode's QUE for the MAC, that use SMPTE time, absolute from
    the beginnning of the track, rather than bar/beat/tick timestamping.
    
    karl