ref: 50791b39057e00de66c38a1856e703f8182b4843
dir: /python/bench-onset/
#! /usr/bin/python
from aubio.bench.config import *
from aubio.bench.node import *
class onset_parameters:
def __init__(self):
""" set default parameters """
self.silence = -70
self.derivate = False
self.localmin = False
self.bufsize = 512
self.hopsize = 256
self.samplerate = 44100
self.tol = 0.05
self.step = float(self.hopsize)/float(self.samplerate)
self.threshold = 0.1
self.mode = 'dual'
class benchonset(bench):
def compute_results(self):
self.P = 100*float(self.expc-self.missed-self.merged)/(self.expc-self.missed-self.merged + self.bad+self.doubled)
self.R = 100*float(self.expc-self.missed-self.merged)/(self.expc-self.missed-self.merged + self.missed+self.merged)
if self.R < 0: self.R = 0
self.F = 2* self.P*self.R / (self.P+self.R)
self.values = [self.params.mode,
"%2.3f" % self.params.threshold,
self.orig,
self.expc,
self.missed,
self.merged,
self.bad,
self.doubled,
(self.orig-self.missed-self.merged),
"%2.3f" % (100*float(self.orig-self.missed-self.merged)/(self.orig)),
"%2.3f" % (100*float(self.bad+self.doubled)/(self.orig)),
"%2.3f" % (100*float(self.orig-self.missed)/(self.orig)),
"%2.3f" % (100*float(self.bad)/(self.orig)),
"%2.3f" % self.P,
"%2.3f" % self.R,
"%2.3f" % self.F ]
def compute_onset(self,input,output):
from aubio.tasks import getonsets, get_onset_mode
from aubio.onsetcompare import onset_roc, onset_diffs
from aubio.txtfile import read_datafile
amode = 'roc'
vmode = 'verbose'
vmode = ''
lres, ofunc = getonsets(input,
self.params.threshold,
self.params.silence,
mode=get_onset_mode(self.params.mode),
localmin=self.params.localmin,
derivate=self.params.derivate,
bufsize=self.params.bufsize,
hopsize=self.params.hopsize,
storefunc=False)
for i in range(len(lres)): lres[i] = lres[i]*self.params.step
ltru = read_datafile(input.replace('.wav','.txt'),depth=0)
if vmode=='verbose':
print "Running with mode %s" % self.params.mode,
print " and threshold %f" % self.params.threshold,
print " on file", input
#print ltru; print lres
if amode == 'localisation':
l = onset_diffs(ltru,lres,self.params.tol)
mean = 0
for i in l: mean += i
if len(l): print "%.3f" % (mean/len(l))
else: print "?0"
elif amode == 'roc':
orig, missed, merged, expc, bad, doubled = onset_roc(ltru,lres,self.params.tol)
self.orig += orig
self.missed += missed
self.merged += merged
self.expc += expc
self.bad += bad
self.doubled += doubled
self.compute_results()
def compute_data(self):
self.orig, self.missed, self.merged, self.expc, \
self.bad, self.doubled = 0, 0, 0, 0, 0, 0
act_on_data(self.compute_onset,self.datadir,self.resdir, \
suffix='',filter='f -name \'*.wav\'')
def run_bench(self,modes=['dual'],thresholds=[0.5]):
self.modes = modes
self.thresholds = thresholds
self.pretty_print(self.titles)
for mode in self.modes:
self.params.mode = mode
for threshold in self.thresholds:
self.params.threshold = threshold
self.compute_data()
self.compute_results()
self.pretty_print(self.values)
def auto_learn(self,modes=['dual'],thresholds=[0.1,1.5]):
""" simple dichotomia like algorithm to optimise threshold """
self.modes = modes
self.pretty_print(self.titles)
for mode in self.modes:
steps = 10
lesst = thresholds[0]
topt = thresholds[1]
self.params.mode = mode
self.params.threshold = topt
self.compute_data()
self.pretty_print(self.values)
topF = self.F
self.params.threshold = lesst
self.compute_data()
self.pretty_print(self.values)
lessF = self.F
for i in range(steps):
self.params.threshold = ( lesst + topt ) * .5
self.compute_data()
self.pretty_print(self.values)
if self.F == 100.0 or self.F == topF:
print "assuming we converged, stopping"
break
#elif abs(self.F - topF) < 0.01 :
# print "done converging"
# break
if topF < self.F:
#lessF = topF
#lesst = topt
topF = self.F
topt = self.params.threshold
elif lessF < self.F:
lessF = self.F
lesst = self.params.threshold
if topt == lesst:
lesst /= 2.
#modes = [ 'complex' ]
modes = ['complex', 'energy', 'phase', 'specdiff', 'kl', 'mkl', 'dual']
#thresholds = [1.5]
thresholds = [ 0.01, 0.05, 0.1, 0.2, 0.3, 0.4, 0.5]
#datapath = "%s%s" % (DATADIR,'/onset/DB/*/')
datapath = "%s%s" % (DATADIR,'/onset/DB/PercussivePhrases/RobertRich')
respath = '/var/tmp/DB-testings'
benchonset = benchonset(datapath,respath,checkres=True,checkanno=True)
benchonset.params = onset_parameters()
benchonset.titles = [ 'mode', 'thres', 'orig', 'expc', 'missd', 'mergd',
'bad', 'doubl', 'corrt', 'GD', 'FP', 'GD-merged', 'FP-pruned',
'prec', 'recl', 'dist' ]
benchonset.formats = ["%12s" , "| %6s", "| %6s", "| %6s", "| %6s", "| %6s",
"| %6s", "| %6s", "| %6s", "| %8s", "| %8s", "| %8s", "| %8s",
"| %6s", "| %6s", "| %6s"]
#benchonset.run_bench(modes=modes,thresholds=thresholds)
benchonset.auto_learn(modes=modes)
# gatherdata
#act_on_data(my_print,datapath,respath,suffix='.txt',filter='f -name \'*.wav\'')
# gatherthreshold
# gathermodes
# comparediffs
# gatherdiffs