annotate testdata/scripts/evaluate_lab.yeti @ 372:af71cbdab621 tip

Update bqvec code

// Take two lab files of the form
//
// onset offset frequency
//
// and report their onset-only note-level accuracy F-measure.

program evaluate_lab;

usage () =
   (eprintln "\nUsage: evaluate_lab reqd reference.lab transcribed.lab\n";
    eprintln "where reqd is the number of milliseconds allowed for timing error (+/-)\n");

toMIDIPitch f = 
    round (12 * (Math#log(f / 220) / Math#log(2)) + 57);

suck f =
   (str = openInFile f "UTF-8"; // better to use readFile here, oh well
    d = map do line:
        case list (strSplit "\t" line) of
        onset::offset::frequency::_:
            { onset = number onset, midi = toMIDIPitch (number frequency) };
        _:
            failWith "badly formed line: \(line)";
        esac;
    done (str.lines ());
    str.close ();
    d);

select f = fold do r x: if f x then x::r else r fi done [];

evaluate permitted ref trans =
   (reference = suck ref;
    transcribed = suck trans;
    accurate = 
        select do here :
            any do other:
                here.midi == other.midi and
                    abs (here.onset - other.onset) < permitted
            done reference
        done transcribed;
    {
        ntot = length transcribed,
        nref = length reference,
        ncorr = length accurate,
    });

pc n =
    int (n * 1000) / 10;

report { ntot, nref, ncorr } =
   (nfp = ntot - ncorr;
    nfn = nref - ncorr;
    if nref == 0 then
        println "ERROR: no events in reference!"
    elif ntot == 0 then
        println "WARNING: no events transcribed!"
    else
        rec = ncorr / nref;
        pre = ncorr / ntot;
        f = if pre + rec == 0 then 0 else 2 * ((pre * rec) / (pre + rec)) fi;
        acc = ncorr / (ncorr + nfp + nfn);
        println "precision \(pc pre), recall \(pc rec), accuracy \(pc acc), F \(pc f)";
    fi);

case (list _argv) of
reqd::ref::trans::[]: report (evaluate (number reqd / 1000) ref trans);
_: usage ();
esac;