Conference napalm::commusic_v1

    I don't own a DX, and don't know about the 0-8 settings and how
    they map into transmitted velocity, but I do own a KX88.
    
    First, if you have a note with a transmitted velocity of 0 it won't
    sound. The MIDI spec allows a range from 0 - 127. The KX88 transmits
    a continuous range from 0 to about 118. Unfortunately. This limited
    range keeps certain instruments from achieving optimum brightness,
    when 'brightness' is determined by incoming MIDI velocity.
    
    karl moeller

    I don't have my DX with me, but (from memory) it has a 0-7 velocity
    sensitivity per oscillator.  I don't know how that maps into MIDI
    velocity, but I recall from looking at a monitor that it has lots
    more than eight velocity values.
        John Sauter

    Gee, I just happen to have my DX with me....;^) 
    
    But no analysis software.
    
    Anyhow, the dx has 8 discrete velocity values, 0-7. From some 
    experiments I did just Sunday night, it appears that increasing
    or decreasing the velocity parameter for any particular oscillator
    increases or decreases the *range* of values to which it will
    respond.
    
    Simplistically, you might say that for every value you change the
    oscillator by, you also expand or contract the range on either
    side of 64 (I think that's default velocity) by 4, although I
    don't think the algorythm is quite that simple.
    
                          0  No range   (all 64)
                          .
                          .
                          7  Full range (all 128)
    
    Edd

    midi spec: vel values 0 to 127, vel 0 == note OFF (!)
    
    possible complications to edd's scenario:
    
    the transfer curve (or 'taper', like on a pot) might be anyshape,
    (read: whatever the mfgr wants. The midi spec suggests a log, I
    bet the easiest firmware is linear.) AND the method of sensing
    velocity (hardware/software/mix) is gonna come into play. It
    may be able to measure 127 different values, or something not
    quite as good (but you claim you can play 126 easy?)
    
    Sooooo....beyond that we could also have:
    
    'sensitivity parameters' like the 0-7 values mentioned, may expand
    the RANGE around some mid point (thanks edd...) OR they may select
    a maximum point, like if you bang it, value 7 yeilds 127max, value
    3 yeilds 64 max....get it? AND there may be associated with that
    some transfer curve that interpolates midi vel values based on the
    max point, and/or any of the above....easy, eh?
    
    ron
    

    Some synths (e.g. The Kurzweil 150) can be adjusted to various tapers,
    so you can use your favorite Midi controller with less fear.
    
    	- Rick
    

    Roland seems to be a little schizoid on this matter - the JX10 has a
    4 position switch for each velocity sensitive parameter that selects
    between "off", linear, and two concave upward curves that have
    different lowest values and slightly different average slopes. 
    These switches thus combine both the "taper" and "sensitivity"
    aspects.  The MKS-80, however, simply has on/off switches for the
    velocity sensitive parameters, and a master sensitivity control
    that affects them all.  There is no "taper" control as such.  I'd
    prefer a combination of these approaches - individual controls per
    affected parameter plus a single master sensitivity control.
    
    In both cases, these parameters are properties of the synth modules.
    
    Now, the Octapad (a controller) has "curve" (taper), sensitivity, and
    "floor (lowest velocity transmitted no matter how softly you hit the thing)
    parameters.  This makes sense for this kind of controller because
    most velocity sensitive drum synths/machines do not offer any control
    over their velocity response.
    
    len.
    

    re 0
    
    I seem to recall that the velocity sensitivity on the DX7 hs nothing
    to do with the keyboard, but rather with the voice.  I did a study
    on the velocity distribution from the keyboard, and came up with
    a distribution of 20 or so values, tightly clustered at the low
    end but not at the high (100,109,118).
    
    David
    

    re DX7 velocity: I recall an article stating that the DX7 can NOT
    send velocity which exceeds 118, even to its own patches. The KX88
    is also somewhat sluggish and 'rolls off' around there as well.