parser.add_argument('-o',
'--outfolder',
type=str,
help='The output folder for sliced audio')
args = parser.parse_args()
if not args.outfolder or not args.infolder:
print('Please provide a valid input/output folder!')
exit()
# Prepare for work
tmp_dir = tempfile.mkdtemp()
input_folder = args.infolder
input_files = os.listdir(input_folder)
output_folder = args.outfolder
check_make(output_folder)
def explode(idx: int):
## NoveltySlice on File
novelty_src = os.path.join(input_folder, input_files[idx])
novelty_indices = os.path.join(tmp_dir, f'{input_files[idx]}_slices.wav')
subprocess.call([
'fluid-noveltyslice', '-source', novelty_src, '-indices',
novelty_indices, '-fftsettings', '2048', '1024', '2048', '-threshold',
'0.61', '-kernelsize', '3', '-filtersize', '1'
])
# Turn slices wav into a list
data = bufspill(novelty_indices)
sound = bufspill(novelty_src)
book_end = len(
# Make paths/dirs
x = x.replace(' ', '_')
# get rid of dots at the start of a file
if x[0] == '.':
x = x[1:]
full_path = os.path.join(root, x)
raw_path = os.path.join(tmp_dir, f'{x}.raw')
audio_path = os.path.join(output_directory, f'{x}.wav')
# Do checks
if current_size < MB_LIM:
if check_size(full_path, 30000) and check_ext(
full_path, BAD_EXTS):
copyfile(full_path, raw_path)
current_size += bytes_to_mb(os.path.getsize(raw_path))
if current_size < MB_LIM:
convert(infile=raw_path,
outfile=audio_path,
encoding=args.encoding,
bits=args.bits,
channels=args.channels)
tmp_dir = tempfile.mkdtemp()
origin_dir = args.infolder
audio_dir = os.path.join(project_root, args.outfolder)
check_make(audio_dir)
scrape(origin_dir, audio_dir)
rmtree(tmp_dir)
printp('Completed Scraping')
import subprocess as sp
import multiprocessing as mp
import os
import sys
import time
from datamosh.utils import check_make, wipe_dir
from datamosh.variables import unique_audio_folder, unique_audio_files, project_root, essentia_analysis
num_jobs = len(unique_audio_files)
simple_analysis_folder = os.path.join(essentia_analysis,
'simple_extractor_segmented')
check_make(simple_analysis_folder)
wipe_dir(simple_analysis_folder)
def process(idx):
input_file = os.path.join(unique_audio_folder, unique_audio_files[idx])
output_file = os.path.join(simple_analysis_folder,
f'{unique_audio_files[idx]}.json')
if not os.path.isfile(output_file):
sp.call([
'essentia_databend_simple_extractor',
input_file,
output_file,
'8192',
'256',
])
with mp.Pool() as pool:
this_script = os.getcwd()
# Configuration
printp('Reading configuration')
cfg_path = os.path.join(this_script, sys.argv[1])
cfg = read_yaml(cfg_path)
json_out = cfg['json']
input_data = cfg['input_data']
algorithm = cfg['algorithm']
normalisation = cfg['normalisation']
identifier = cfg['identifier']
folder_name = f'{algorithm}_{identifier}'
output_path = os.path.join(this_script, 'outputs', folder_name)
check_make(output_path)
copyfile(cfg_path, os.path.join(output_path, 'configuration.yaml'))
printp('Reading in data')
feature = read_json(os.path.join(project_root, 'python_scripts', 'dimensionality_reduction', 'outputs', input_data))
keys = [x for x in feature.keys()]
values = [y for y in feature.values()]
data = np.array(values)
printp('Normalising')
if normalisation != 'none':
if normalisation == 'minmax':
scaler = MinMaxScaler()
if normalisation == 'standardise':
scaler = StandardScaler()
import subprocess as sp
import os
import sys
import multiprocessing as mp
from datamosh.variables import unique_audio_folder, unique_audio_files, project_root
from datamosh.utils import check_make
filtered_folder = os.path.join(project_root, 'DataAudioUnique_DC')
check_make(filtered_folder)
num_jobs = len(unique_audio_files)
def remove_dc(infile: str, outfile: str, hertz: str):
sp.Popen(['sox', '-D', '-V0', infile, outfile, 'highpass', hertz])
def process(idx: int):
remove_dc(os.path.join(unique_audio_folder, unique_audio_files[idx]),
os.path.join(filtered_folder, unique_audio_files[idx]), '10')
with mp.Pool() as pool:
for i, _ in enumerate(pool.imap_unordered(process, range(num_jobs)), 1):
sys.stderr.write('\rdone {0:%}'.format(i / num_jobs))