def roughness_curve(
ref_chord: ChordSpectrum,
test_chord: ChordSpectrum,
# ref_chord_struct: list = de.default_chord_struct,
# chord_struct_type: str = de.default_chord_struct_type,
*,
# fund_hz: float = de.default_fund,
# ref_timbre: pd.DataFrame = de.default_timbre,
# test_chord_struct: list = de.default_chord_struct,
# test_timbre: pd.DataFrame = de.default_timbre,
transpose_domain: TransposeDomain = de.default_transpose_domain,
function_type: str = de.default_roughness_function_type,
normalize: bool = False,
plot: bool = False,
options: Dict = {
'crossterms_only': False,
'amp_type': 'MIN',
'cutoff': False,
'original': False,
'show_partials': False
}
) -> ArrayLike:
# Using Sethares' original function. Note that incorporating crossterms only
# (i.e. interactions between the two chords, not roughness relations of
# partials within chord) removes register-dependent effects due to
# self-roughness alone.
#if options['original']:
# options['crossterms_only'] = False
# ref_chord = cu.make_chord(ref_chord_struct, chord_struct_type, timbre=ref_timbre, fund_hz=fund_hz)
# new_test_timbre = test_timbre.copy()
if (ref_chord == test_chord):
copy_tim = Timbre(ref_chord.partials['hz_orig'],
ref_chord.partials['amp'])
test_chord = ChordSpectrum([0], 'ST_DIFF', timbre=copy_tim, fund_hz=1)
min_hz = np.min(test_chord.partials['hz_orig'])
test_chord.partials['fund_multiple'] /= min_hz
roughness_vals = np.zeros(np.shape(transpose_domain.domain))
# if chord_struct_type.upper() == 'HZ_SHIFT':
# fund_hz = 0
if options['crossterms_only']:
if options['show_partials']:
ref_self_diss = (roughness_complex(ref_chord,
function_type,
options=options))['roughness']
else:
ref_self_diss = (roughness_complex(ref_chord,
function_type,
options=options))
for (idx, position) in enumerate(transpose_domain.domain):
# new_test_timbre['fund_multiple'] = cu.slide_timbre(position, test_timbre, chord_struct_type=chord_struct_type)
# test_chord = cu.make_chord(test_chord_struct, chord_struct_type, timbre=new_test_timbre, fund_hz=fund_hz)
test_chord.transpose(position, transpose_domain.transpose_type)
# union = ref_chord.append(test_chord, ignore_index=True)
if function_type.upper() == 'HELMHOLTZ':
union = MergedSpectrum(test_chord)
else:
union = MergedSpectrum(ref_chord, test_chord)
if options['show_partials']:
curr_roughness_val = (roughness_complex(
union, function_type, options=options))['roughness']
else:
curr_roughness_val = (roughness_complex(union,
function_type,
options=options))
if options['crossterms_only']:
if options['show_partials']:
test_self_diss = (roughness_complex(
ref_chord, function_type, options=options))['roughness']
else:
test_self_diss = (roughness_complex(ref_chord,
function_type,
options=options))
curr_roughness_val -= (ref_self_diss + test_self_diss)
roughness_vals[idx] = curr_roughness_val
# test_chord has been mutated by .transpose(); need to reset
test_chord.reset_partials()
if normalize:
plotMax = max(roughness_vals)
roughness_vals /= float(plotMax)
if plot:
plt.plot(transpose_domain.domain, roughness_vals)
plt.show()
return roughness_vals
def overlap_curve(
ref_chord: ChordSpectrum,
test_chord: ChordSpectrum,
*,
transpose_domain: TransposeDomain = de.default_transpose_domain,
function_type: str = de.default_overlap_function_type,
normalize: bool = False,
options: Dict = {
'crossterms_only': False,
'amp_type': 'MIN',
'cutoff': False,
'original': False,
'show_partials': False
}
) -> ArrayLike:
overlap_vals = np.zeros(np.shape(transpose_domain.domain))
if options['crossterms_only']:
if options['show_partials']:
ref_self_overlap = (overlap_complex(ref_chord,
function_type,
options=options))['overlap']
else:
ref_self_overlap = (overlap_complex(ref_chord,
function_type,
options=options))
for (idx, position) in enumerate(transpose_domain.domain):
# new_test_timbre['fund_multiple'] = cu.slide_timbre(position, test_timbre, chord_struct_type=chord_struct_type)
# test_chord = cu.make_chord(test_chord_struct, chord_struct_type, timbre=new_test_timbre, fund_hz=fund_hz)
test_chord.transpose(position, transpose_domain.transpose_type)
# union = ref_chord.append(test_chord, ignore_index=True)
union = MergedSpectrum(ref_chord, test_chord)
if options['show_partials']:
curr_overlap_val = (overlap_complex(union,
function_type,
options=options))['overlap']
else:
curr_overlap_val = (overlap_complex(union,
function_type,
options=options))
if options['crossterms_only']:
if options['show_partials']:
test_self_overlap = (overlap_complex(
ref_chord, function_type, options=options))['overlap']
else:
test_self_overlap = (overlap_complex(ref_chord,
function_type,
options=options))
curr_overlap_val -= (ref_self_overlap + test_self_overlap)
overlap_vals[idx] = curr_overlap_val
# test_chord has been mutated by .transpose(); need to reset
test_chord.reset_partials()
if normalize:
plotMax = max(overlap_vals)
overlap_vals /= float(plotMax)
return overlap_vals