def parse(filename):
"""Decode a filename matching the UIowa format.
Parameters
----------
filename : str
A filename conforming to the UIowa MIS conventions.
Returns
-------
instrument : str, or None
Instrument name
dynamic : str
Dymanic marker
notevalue : str
Note names for the file.
"""
filename = utils.filebase(filename)
parameters = [x.strip() for x in filename.split('.')]
instrument = parameters.pop(0)
# This regex matches note names with a preceeding and following '.'
note_match = re.search(r"(?<=\.)[A-Fb#0-6]*(?<!\.)", filename)
notevalue = filename[note_match.start():note_match.end()] \
if note_match else None
# This regex matches dynamic chars with a preceeding and following '.'
dynamic_match = re.search(r"(?<=\.)[f|p|m]*(?<!\.)", filename)
dynamic = filename[dynamic_match.start():dynamic_match.end()] \
if dynamic_match else None
return instrument, dynamic, notevalue
def parse(filename):
"""Decode a filename matching the UIowa format.
Parameters
----------
filename : str
A filename conforming to the UIowa MIS conventions.
Returns
-------
instrument : str, or None
Instrument name
dynamic : str
Dymanic marker
notevalue : str
Note names for the file.
"""
filename = utils.filebase(filename)
parameters = [x.strip() for x in filename.split('.')]
instrument = parameters.pop(0)
# This regex matches note names with a preceeding and following '.'
note_match = re.search(r"(?<=\.)[A-Fb#0-6]*(?<!\.)", filename)
notevalue = filename[note_match.start():note_match.end()] \
if note_match else None
# This regex matches dynamic chars with a preceeding and following '.'
dynamic_match = re.search(r"(?<=\.)[f|p|m]*(?<!\.)", filename)
dynamic = filename[dynamic_match.start():dynamic_match.end()] \
if dynamic_match else None
return instrument, dynamic, notevalue
def parse(filename):
"""Convert phil path to codes/parameters.
Parameters
----------
filename : full path.
Returns
-------
parts : tuple, len=5
From the filename, the following parts:
(instrument, note, duration, dynamic, articulation).
"""
audio_file_name = utils.filebase(filename)
(instrument, note, duration, dynamic,
articulation) = audio_file_name.split('_')
return instrument, note, duration, dynamic, articulation
def parse(filename):
"""Convert phil path to codes/parameters.
Parameters
----------
filename : full path.
Returns
-------
parts : tuple, len=5
From the filename, the following parts:
(instrument, note, duration, dynamic, articulation).
"""
audio_file_name = utils.filebase(filename)
(instrument, note, duration, dynamic,
articulation) = audio_file_name.split('_')
return instrument, note, duration, dynamic, articulation
def parse(filename):
"""Convert good-sounds path to codes/parameters.
Parameters
----------
filename : full path.
Returns
-------
parts : tuple, len=5
From the filename, the following parts:
(instrument, pack, take, note_id)
"""
if EXPECTED_ROOT_DIR in filename:
filename = filename[filename.index(EXPECTED_ROOT_DIR) +
len(EXPECTED_ROOT_DIR):]
(instrument, pack, take, note_id) = (filename.strip('/').split('/'))
note_id = utils.filebase(note_id)
return instrument, pack, take, note_id
def collect(base_dir, fext="*.flac", onset_dir=ONSET_DIR):
"""Convert a base directory of RWC files to a pa datafra
Parameters
----------
base_dir : str
Full path to the base RWC directory.
dataset : str
Datset string to use in this df.
Returns
-------
pandas.DataFrame
Indexed by:
id : [dataset identifier] + [8 char md5 of filename]
With the following columns:
audio_file : full path to original audio file.
dataset : dataset it is from
instrument : instrument label.
dynamic : dynamic tag
"""
logger.info("Scanning {} for audio files.".format(base_dir))
indexes = []
records = []
fmt = "*/*/{}".format(fext)
for audio_file_path in glob.glob(os.path.join(base_dir, fmt)):
instrument_name, style_code, dynamic_code = parse(audio_file_path)
# TODO: Other datasets hash the filepath relative to `base_dir`; we
# should really do the same here, but care must be taken to keep the
# onsets sync'ed.
uid = utils.generate_id(NAME, utils.filebase(audio_file_path))
onsets = utils.find_onset_file_from_uid(uid, onset_dir)
indexes.append(uid)
records.append(
dict(audio_file=audio_file_path,
dataset=NAME,
instrument=instrument_name,
dynamic=dynamic_code,
onsets_file=onsets))
logger.info("Found {} files from RWC.".format(len(records)))
return pd.DataFrame(records, index=indexes)
def collect(base_dir, fext="*.flac", onset_dir=ONSET_DIR):
"""Convert a base directory of RWC files to a pa datafra
Parameters
----------
base_dir : str
Full path to the base RWC directory.
dataset : str
Datset string to use in this df.
Returns
-------
pandas.DataFrame
Indexed by:
id : [dataset identifier] + [8 char md5 of filename]
With the following columns:
audio_file : full path to original audio file.
dataset : dataset it is from
instrument : instrument label.
dynamic : dynamic tag
"""
logger.info("Scanning {} for audio files.".format(base_dir))
indexes = []
records = []
fmt = "*/*/{}".format(fext)
for audio_file_path in glob.glob(os.path.join(base_dir, fmt)):
instrument_name, style_code, dynamic_code = parse(audio_file_path)
# TODO: Other datasets hash the filepath relative to `base_dir`; we
# should really do the same here, but care must be taken to keep the
# onsets sync'ed.
uid = utils.generate_id(NAME, utils.filebase(audio_file_path))
onsets = utils.find_onset_file_from_uid(uid, onset_dir)
indexes.append(uid)
records.append(
dict(audio_file=audio_file_path,
dataset=NAME,
instrument=instrument_name,
dynamic=dynamic_code,
onsets_file=onsets))
logger.info("Found {} files from RWC.".format(len(records)))
return pd.DataFrame(records, index=indexes)
def parse(filename):
"""Convert good-sounds path to codes/parameters.
Parameters
----------
filename : full path.
Returns
-------
parts : tuple, len=5
From the filename, the following parts:
(instrument, pack, take, note_id)
"""
if EXPECTED_ROOT_DIR in filename:
filename = filename[filename.index(EXPECTED_ROOT_DIR) +
len(EXPECTED_ROOT_DIR):]
(instrument, pack, take, note_id) = (
filename.strip('/').split('/'))
note_id = utils.filebase(note_id)
return instrument, pack, take, note_id
def collect(base_dir, fext="*.flac"):
"""Convert a base directory of RWC files to a pandas dataframe.
Parameters
----------
base_dir : str
Full path to the base RWC directory.
dataset : str
Datset string to use in this df.
Returns
-------
pandas.DataFrame
Indexed by:
id : [dataset identifier] + [8 char md5 of filename]
With the following columns:
audio_file : full path to original audio file.
dataset : dataset it is from
instrument : instrument label.
dynamic : dynamic tag
"""
logger.info("Scanning {} for audio files.".format(base_dir))
indexes = []
records = []
fmt = "*/*/{}".format(fext)
for audio_file_path in glob.glob(os.path.join(base_dir, fmt)):
instrument_name, style_code, dynamic_code = parse(audio_file_path)
uid = utils.generate_id(NAME, utils.filebase(audio_file_path))
indexes.append(uid)
records.append(
dict(audio_file=audio_file_path,
dataset=NAME,
instrument=instrument_name,
dynamic=dynamic_code))
logger.info("Found {} files from RWC.".format(len(records)))
return pd.DataFrame(records, index=indexes)
def parse(filename):
"""Takes an rwc path, and returns the extracted codes from the
filename.
Parameters
----------
rwc_path : str
Full path or basename. If full path, gets the basename.
Returns
-------
instrument_name : str, or None if cannot be parsed.
style_code : str
dynamic_code : str
"""
filebase = utils.filebase(filename)
instrument_code = filebase[3:5]
# Get the instrument name from the json file.
instrument_name = instrument_code_to_name(instrument_code)
style_code = filebase[5:7]
dynamic_code = filebase[7]
return instrument_name, style_code, dynamic_code
def parse(filename):
"""Takes an rwc path, and returns the extracted codes from the
filename.
Parameters
----------
rwc_path : str
Full path or basename. If full path, gets the basename.
Returns
-------
instrument_name : str, or None if cannot be parsed.
style_code : str
dynamic_code : str
"""
filebase = utils.filebase(filename)
instrument_code = filebase[3:5]
# Get the instrument name from the json file.
instrument_name = instrument_code_to_name(instrument_code)
style_code = filebase[5:7]
dynamic_code = filebase[7]
return instrument_name, style_code, dynamic_code