def onset_detection(input_file, output_file, threshold, freqs, log_flag):
'''Onset detection
:parameters:
- input_file : str
path to input audio file (wav, mp3, m4a, flac, etc.)
- annotations_file : str
path to the annotations file (txt, csv, etc)
- delimiter: str
delimiter string to process the annotations file
'''
# 1. load the wav file
# use an example audio file provided
print('Loading audio file ...', input_file)
y, sr = audio.load(input_file, sr=None)
# 2. values for compute accentuation feature
hop = 5e-3 # hop size
nfilts = 80 # Number of MEL filters
alpha = 10e4 # compression parameter for dB conversion - log10(alpha*abs(S)+1)
# 3. values for peak detection on the accentuation feature function
pre_avg = 14 # number of past frames for moving average
pos_avg = 10 # number of future frames for moving average
pre_max = 14 # number of past frames for moving maximum
pos_max = 10 # number of future frames for moving maximum
# 4. onsets detection
print('Detecting onsets')
onset_times, feature_values = onsets.detection(y,
fs=sr,
hop=hop,
nfilts=nfilts,
log_flag=log_flag,
alpha=alpha,
minfreq=freqs[0],
maxfreq=freqs[1],
threshold=threshold,
pre_avg=pre_avg,
pos_avg=pos_avg,
pre_max=pre_max,
pos_max=pos_max)
# 5. plot everything
plt.plot()
display.wave_plot(y, sr, onsets=onset_times)
plt.title('detected onsets')
plt.show()
# 6. save onsets to csv
print('Saving onset annotations ...', output_file)
annotations.save_onsets(output_file, onset_times)
def onset_detection(input_file, output_file, preset_name, json_file, verbose):
'''Onset detection
:parameters:
- input_file : str
path to input audio file (wav, mp3, m4a, flac, etc.)
- output_file : str
path to the output csv file
- preset_name : str
name of the preset to load (e.g. "chico")
- json_file : str
path to the json file of presets. Default: carat/presets/onsets.json
- verbose : bool
if True print values of parameters in chosen preset. Default: True
'''
# 1. load the wav file
# use an example audio file provided
print('Loading audio file ...', input_file)
y, sr = audio.load(input_file, sr=None)
# 2. load preset values
print('Loading preset parameter values from', json_file, 'json file...')
preset = util.load_preset(preset_name, json_file)
if verbose:
print('Values in', preset_name, 'preset are:', preset)
# 3. onsets detection
print('Detecting onsets')
onset_times, feature_values = onsets.detection(y, fs=sr, **preset)
# 4. plot everything
plt.plot()
display.wave_plot(y, sr, onsets=onset_times)
plt.title('detected onsets')
plt.show()
# 5. save onsets to csv
print('Saving onset annotations ...', output_file)
annotations.save_onsets(output_file, onset_times)
def align_beats_to_onsets(audio_file, beat_annotations_file,
onset_annotations_file, delimiter):
'''Align the location of beats to the location of onsets closest to the beats
:parameters:
- audio_file : str
path to input audio file (wav, mp3, m4a, flac, etc.)
- beat_annotations_file : str
path to the beat annotations file (txt, csv, etc)
- onset_annotations_file : str
path to the annotations file (txt, csv, etc)
- delimiter: str
delimiter string to process the annotations file
'''
# 1. load the wav file
print('Loading audio file ...', audio_file)
y, sr = audio.load(audio_file, sr=None, duration=30.0)
# 2. load beat annotations
print('Loading beat annotations ...', beat_annotations_file)
beats, _ = annotations.load_beats(beat_annotations_file,
delimiter=delimiter)
# 3. load onset annotations
print('Loading onset annotations ...', onset_annotations_file)
onsets, _ = annotations.load_onsets(onset_annotations_file,
delimiter=delimiter)
# 4. compute beats from onsets
print('Computing beats from onsets ...')
beat_ons = microtiming.beats_from_onsets(beats, onsets)
# 5. plot everything
ax1 = plt.subplot(2, 1, 1)
display.wave_plot(y, sr, ax=ax1, beats=beats)
# plot aligned beats
ax2 = plt.subplot(2, 1, 2, sharex=ax1)
display.wave_plot(y, sr, ax=ax2, beats=beat_ons)
plt.show()
# within the rhythm cycle (e.g. 1.1, 1.2, or 1, 2). # # **Note 3:** The same annotations file is used for both beats and downbeats. # This is based on annotation labels that provide a particular string to identify the downbeats. # In this case, this string is .1, and is the one used by default. You can specify the string to # look for in the labels data to select downbeats by setting the `downbeat_label` parameter value. # For instance, `downbeat_label='1'` is used for loading annotations of the samba files included. # # **Note 4:** By default the columns are assumed to be separated by a comma, but you can specify # another separating string by setting the `delimiter` parameter value. For instance, a blank space # `delimiter=' '` is used for loading annotations of the samba files included. # # Let's print the first 10 beat and the first 3 downbeats, with their corresponding labels. print(beats[:10]) print(beat_labs[:10]) print(downbeats[:3]) print(downbeat_labs[:3]) ############################################## # Finally we plot the audio waveform and the beat annotations plt.figure(figsize=(12, 6)) ax1 = plt.subplot(2, 1, 1) display.wave_plot(y, sr, ax=ax1) ax2 = plt.subplot(2, 1, 2, sharex=ax1) display.wave_plot(y, sr, ax=ax2, beats=downbeats, beat_labs=downbeat_labs) plt.tight_layout() plt.show()
# Next, we'll load the annotations provided for the example audio file.
# We get the path to the annotations file corresponding to example number 1,
# and then we load beats and downbeats, along with their labels.
annotations_path = util.example("ansina_beats")
beats, beat_labs = annotations.load_beats(annotations_path)
downbeats, downbeat_labs = annotations.load_downbeats(annotations_path)
##############################################
# Then, we'll compute the accentuation feature.
#
# **Note:** This example is tailored towards the rhythmic patterns of the lowest
# sounding of the three drum types taking part in the recording, so the analysis
# focuses on the low frequencies (20 to 200 Hz).
acce, times, _ = features.accentuation_feature(y, sr, minfreq=20, maxfreq=200)
##############################################
# Finally we plot the audio waveform, the beat annotations and the accentuation feature values.
# plot waveform and accentuation feature
plt.figure(figsize=(12, 6))
# plot waveform
ax1 = plt.subplot(2, 1, 1)
display.wave_plot(y, sr, ax=ax1, beats=beats, beat_labs=beat_labs)
# plot accentuation feature
ax2 = plt.subplot(2, 1, 2, sharex=ax1)
display.feature_plot(acce, times, ax=ax2, beats=beats, beat_labs=beat_labs)
plt.tight_layout()
plt.show()