Skip to content

Commit

Permalink
incorporated comments from @dpwe
Browse files Browse the repository at this point in the history
  • Loading branch information
@bmcfee
bmcfee committed May 8, 2016
1 parent 1cdfcfa commit 3a1ef3b
Showing 1 changed file with 40 additions and 52 deletions.
92 changes: 40 additions & 52 deletions librosa/core/constantq.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -3,9 +3,9 @@
'''Pitch-tracking and tuning estimation'''
from __future__ import division

from warnings import warn
import numpy as np
import scipy.fftpack as fft
from warnings import warn

from . import audio
from .time_frequency import cqt_frequencies, note_to_hz
Expand Down Expand Up @@ -193,18 +193,16 @@ def cqt(y, sr=22050, hop_length=512, fmin=None, n_bins=84,
if res_type != 'kaiser_fast':

# Do the top octave before resampling to allow for fast resampling
fft_basis, n_fft, _ = __fft_filters(sr, fmin_t,
n_filters,
bins_per_octave,
tuning,
filter_scale,
norm,
sparsity)

my_cqt = __fft_response(y, n_fft, hop_length, fft_basis)
fft_basis, n_fft, _ = __cqt_filter_fft(sr, fmin_t,
n_filters,
bins_per_octave,
tuning,
filter_scale,
norm,
sparsity)

# Convolve
cqt_resp.append(my_cqt)
# Compute the CQT filter response and append it to the stack
cqt_resp.append(__cqt_response(y, n_fft, hop_length, fft_basis))

fmin_t /= 2
fmax_t /= 2
Expand All @@ -222,13 +220,13 @@ def cqt(y, sr=22050, hop_length=512, fmin=None, n_bins=84,
.format(n_octaves - 1, n_octaves))

# Now do the recursive bit
fft_basis, n_fft, _ = __fft_filters(sr, fmin_t,
n_filters,
bins_per_octave,
tuning,
filter_scale,
norm,
sparsity)
fft_basis, n_fft, _ = __cqt_filter_fft(sr, fmin_t,
n_filters,
bins_per_octave,
tuning,
filter_scale,
norm,
sparsity)

my_y, my_sr, my_hop = y, sr, hop_length

Expand All @@ -248,11 +246,9 @@ def cqt(y, sr=22050, hop_length=512, fmin=None, n_bins=84,
my_sr /= 2.0
my_hop //= 2

# Compute a dynamic hop based on n_fft
my_cqt = __fft_response(my_y, n_fft, my_hop, fft_basis)
# Compute the cqt filter response and append to the stack
cqt_resp.append(__cqt_response(my_y, n_fft, my_hop, fft_basis))

# Convolve
cqt_resp.append(my_cqt)

return __trim_stack(cqt_resp, n_bins, real)

Expand Down Expand Up @@ -361,31 +357,27 @@ def hybrid_cqt(y, sr=22050, hop_length=512, fmin=None, n_bins=84,

if n_bins_pseudo > 0:
fmin_pseudo = np.min(freqs[pseudo_filters])
my_pseudo_cqt = pseudo_cqt(y, sr,
cqt_resp.append(pseudo_cqt(y, sr,
hop_length=hop_length,
fmin=fmin_pseudo,
n_bins=n_bins_pseudo,
bins_per_octave=bins_per_octave,
tuning=tuning,
filter_scale=filter_scale,
norm=norm,
sparsity=sparsity)

cqt_resp.append(my_pseudo_cqt)
sparsity=sparsity))

if n_bins_full > 0:
my_cqt = np.abs(cqt(y, sr,
hop_length=hop_length,
fmin=fmin,
n_bins=n_bins_full,
bins_per_octave=bins_per_octave,
tuning=tuning,
filter_scale=filter_scale,
norm=norm,
sparsity=sparsity,
real=False))

cqt_resp.append(my_cqt)
cqt_resp.append(np.abs(cqt(y, sr,
hop_length=hop_length,
fmin=fmin,
n_bins=n_bins_full,
bins_per_octave=bins_per_octave,
tuning=tuning,
filter_scale=filter_scale,
norm=norm,
sparsity=sparsity,
real=False)))

return __trim_stack(cqt_resp, n_bins, True)

Expand Down Expand Up @@ -468,15 +460,11 @@ def pseudo_cqt(y, sr=22050, hop_length=512, fmin=None, n_bins=84,
if tuning is None:
tuning = estimate_tuning(y=y, sr=sr)

fft_basis, n_fft, _ = __fft_filters(sr,
fmin,
n_bins,
bins_per_octave,
tuning,
filter_scale,
norm,
sparsity,
hop_length=hop_length)
fft_basis, n_fft, _ = __cqt_filter_fft(sr, fmin, n_bins,
bins_per_octave,
tuning, filter_scale,
norm, sparsity,
hop_length=hop_length)

fft_basis = np.abs(fft_basis)

Expand All @@ -487,8 +475,8 @@ def pseudo_cqt(y, sr=22050, hop_length=512, fmin=None, n_bins=84,
return fft_basis.dot(D)


def __fft_filters(sr, fmin, n_bins, bins_per_octave, tuning,
filter_scale, norm, sparsity, hop_length=None):
def __cqt_filter_fft(sr, fmin, n_bins, bins_per_octave, tuning,
filter_scale, norm, sparsity, hop_length=None):
'''Generate the frequency domain constant-Q filter basis.'''

basis, lengths = filters.constant_q(sr,
Expand All @@ -506,7 +494,7 @@ def __fft_filters(sr, fmin, n_bins, bins_per_octave, tuning,
if hop_length is not None and n_fft < 2.0**(1 + np.ceil(np.log2(hop_length))):
n_fft = int(2.0 ** (1 + np.ceil(np.log2(hop_length))))

# normalize by inverse length to compensate for phase invariance
# re-normalize bases with respect to the FFT window length
basis *= lengths[:, np.newaxis] / float(n_fft)

# FFT and retain only the non-negative frequencies
Expand Down Expand Up @@ -535,7 +523,7 @@ def __trim_stack(cqt_resp, n_bins, real):
return C


def __fft_response(y, n_fft, hop_length, fft_basis):
def __cqt_response(y, n_fft, hop_length, fft_basis):
'''Compute the filter response with a target STFT hop.'''

# Compute the STFT matrix
Expand Down

0 comments on commit 3a1ef3b

Please sign in to comment.