@performance_data()

returns a tab, each element of which documents the occurrence of a musical event from the start of the program.

Each event is described by a tab listing the following information:

1. index : the index (rank) of the event in the score (or 0 if a score event is not meaningfull)

2. date: the date of the occurence of this event, expressed in seconds from the start of the program. This date is the value of the $NOW variable at the occurence of the event.

3. beat position : The date in beat of this event (i.e., its 'position' in the score).

4. reset : true if this event specifies a new tempo value. This is the case for the first event following a BPM specifciation. This is also the case when the tempo is manually set (using the antescofo::tempo command): an entry not related to a score event is inserted in the performance data (the index is 0) and reset is true.

5. tempo infered : True if the tempo has been infered. A new tempo value is not computed at the very start of the score, after a jump, after a silence, etc.

6. tempo : the value of the tempo at this event.

7. $\kappa$ : The current $\kappa$ value (a measure of the randomness of the interpretation), see coupling strategies. This parameter is related to the certainty of the computed tempo: the greater its value, the greater the expectation that the current tempo $T$ will not vary. The value of $\kappa$ is inversely related to the width of a unimodal centered on $T$. The larger $\kappa$, the tighter the unimodal is on $T$. For $\kappa = 0$, all tempos in $[T/2, 2T]$ are equiprobable. Parameter $\kappa$ is used to compute the value of the coupling parameters $\eta_{\,p}$ and $\eta_{\,\phi}$ in the coupling strategy.

8. $\eta_{\,p}$ : The parameter $\eta_{\,p}$ used in the tempo inference. Its value is a function of $\kappa$ defined by the coupling strategy currently in use. It value is in $[0, 1]$. The smaller the value, the less the tempo is corrected. Cf. coupling strategies.

9. $\eta_{\,\phi}$ : The parameter $\eta_{\,\phi}$ used in the tempo inference. Its value is a function of $\kappa$ defined by the coupling strategy currently in use. It value is in $[0, 2]$. The smaller the value, the less the phase is corrected. Cf. coupling strategies.

10. $\phi$ : this parameter is the mapping of the durantion of the previous event on the clock that is coupled with the musician. See the Large and Jones inference model in The dynamics of attending: How people track time-varying events. The value belongs to $[-1/2, +1/2]$. A value of $0$ means perfext synchrony, i.e. the antescofo clock is fully synchronized with the musician 'internal' clock. A deviation from 0 measures the approximation of the phase and the speed (tempo) of the musician.

11. missed : True if the previous event is a missed event.

12. duration : True if the previous event has a definite duration (this is not the case for silence, grace note, etc.).

13. This element is reserved for internal use.

14. This element is reserved for internal use.

The information returned by @performance_data()can be used to plot the evolution of the infered tempo. The following fragment can be used to visualize the tempo and th instantaneous tempo.

    $data := @performance_data()

    $t := [ $e[1]  | $e in $data]    // date (in seconds) of the events
    $tempo:= [ $e[5]  | $e in $data] // tempo at each event

    // duration of each event (except the last) shifted to the left
    // i.e., the first tab element is the duration of the second event
    $dur := [ $data[$i, 2] - $data[$i-1, 2]) | $i in 1 .. $data.size() ]

    // instantaneous tempo of each event, except the first
    $itempo := [ 60 * $dur[$i] / ($data[$i, 1] - $data[$i-1, 1]) | $i in 1 .. $data.size() ]

    // date of each event in $itempo
    $tt := $t.drop(1)


    print plot3
    _ := @gnuplot("rtTempo", $t, $tempo, "iTempo", $tt, $itempo)
    _ := @gnuplot("kappa", $t, @min(10, $e[6]) | $e in $data])

A possible output on the Ravel example in the Antescofo distribution package is (open the image in another tab to have a full view):

   

The $\kappa$ plot is clipped at $10$ to show better the small fluctuations.