Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Sanitized quantized filenames #63

Open
wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Sanitized quantized filenames #63

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
65 changes: 35 additions & 30 deletions polymath.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,6 +2,7 @@


import os
import re
import sys
import pickle
import argparse
Expand Down Expand Up @@ -344,26 +345,32 @@ def extractMIDI(audio_paths, output_dir):
save_model_outputs=save_model_outputs,
save_notes=save_notes)

def sanitize_filename(filename):
# Remove any invalid characters
filename = re.sub(r'[<>:"/\\|?*\x00-\x1F]', '_', filename)
# Remove leading/trailing spaces
filename = filename.strip()
return filename

def quantizeAudio(vid, bpm=120, keepOriginalBpm = False, pitchShiftFirst = False, extractMidi = False):
def quantizeAudio(vid, bpm=120, keepOriginalBpm=False, pitchShiftFirst=False, extractMidi=False):
print("Quantize Audio: Target BPM", bpm,
"-- id:",vid.id,
"bpm:",round(vid.audio_features["tempo"],2),
"frequency:",round(vid.audio_features['frequency'],2),
"key:",vid.audio_features['key'],
"timbre:",round(vid.audio_features['timbre'],2),
"name:",vid.name,
'keepOriginalBpm:', keepOriginalBpm
)

# load audio file
"-- id:", vid.id,
"bpm:", round(vid.audio_features["tempo"], 2),
"frequency:", round(vid.audio_features['frequency'], 2),
"key:", vid.audio_features['key'],
"timbre:", round(vid.audio_features['timbre'], 2),
"name:", vid.name,
'keepOriginalBpm:', keepOriginalBpm
)

# Load audio file
y, sr = librosa.load(vid.audio, sr=None)

# Keep Original Song BPM
# Keep original song BPM
if keepOriginalBpm:
bpm = float(vid.audio_features['tempo'])
print('Keep original audio file BPM:', vid.audio_features['tempo'])
# Pitch Shift audio file to desired BPM first
# Pitch shift audio file to desired BPM first
elif pitchShiftFirst: # WORK IN PROGRESS
print('Pitch Shifting audio to desired BPM', bpm)
# Desired tempo in bpm
Expand All @@ -374,70 +381,68 @@ def quantizeAudio(vid, bpm=120, keepOriginalBpm = False, pitchShiftFirst = False
y = librosa.resample(y=y, orig_sr=sr, target_sr=sr_stretched) #, res_type='linear'
y = librosa.resample(y, orig_sr=sr, target_sr=44100)

# extract beat
# Extract beat
y_harmonic, y_percussive = librosa.effects.hpss(y)
tempo, beats = librosa.beat.beat_track(sr=sr, onset_envelope=librosa.onset.onset_strength(y=y_percussive, sr=sr), trim=False)
beat_frames = librosa.frames_to_samples(beats)

# generate metronome
# Generate metronome
fixed_beat_times = []
for i in range(len(beat_frames)):
fixed_beat_times.append(i * 120 / bpm)
fixed_beat_frames = librosa.time_to_samples(fixed_beat_times)

# construct time map
# Construct time map
time_map = []
for i in range(len(beat_frames)):
new_member = (beat_frames[i], fixed_beat_frames[i])
time_map.append(new_member)

# add ending to time map
# Add ending to time map
original_length = len(y+1)
orig_end_diff = original_length - time_map[i][0]
new_ending = int(round(time_map[i][1] + orig_end_diff * (tempo / bpm)))
new_member = (original_length, new_ending)
time_map.append(new_member)

# time strech audio
# Time stretch audio
print('- Quantize Audio: source')
strechedaudio = pyrb.timemap_stretch(y, sr, time_map)

path_suffix = (
path_suffix = sanitize_filename(
f"Key {vid.audio_features['key']} - "
f"Freq {round(vid.audio_features['frequency'], 2)} - "
f"Timbre {round(vid.audio_features['timbre'], 2)} - "
f"BPM Original {int(vid.audio_features['tempo'])} - "
f"BPM {bpm}"
)
path_prefix = (
f"{vid.id} - {vid.name}"
)
path_prefix = sanitize_filename(f"{vid.id} - {vid.name}")

audiofilepaths = []
# save audio to disk
path = os.path.join(os.getcwd(), 'processed', path_prefix + " - " + path_suffix +'.wav')
# Save audio to disk
path = os.path.join(os.getcwd(), 'processed', path_prefix + " - " + path_suffix + '.wav')
sf.write(path, strechedaudio, sr)
audiofilepaths.append(path)

# process stems
# Process stems
stems = ['bass', 'drums', 'guitar', 'other', 'piano', 'vocals']
for stem in stems:
path = os.path.join(os.getcwd(), 'separated', 'htdemucs_6s', vid.id, stem +'.wav')
path = os.path.join(os.getcwd(), 'separated', 'htdemucs_6s', vid.id, sanitize_filename(stem) + '.wav')
print(f"- Quantize Audio: {stem}")
y, sr = librosa.load(path, sr=None)
strechedaudio = pyrb.timemap_stretch(y, sr, time_map)
# save stems to disk
path = os.path.join(os.getcwd(), 'processed', path_prefix + " - Stem " + stem + " - " + path_suffix +'.wav')
# Save stems to disk
path = os.path.join(os.getcwd(), 'processed', path_prefix + " - Stem " + sanitize_filename(stem) + " - " + path_suffix + '.wav')
sf.write(path, strechedaudio, sr)
audiofilepaths.append(path)

# metronome click (optinal)
# Metronome click (optional)
click = False
if click:
clicks_audio = librosa.clicks(times=fixed_beat_times, sr=sr)
print(len(clicks_audio), len(strechedaudio))
clicks_audio = clicks_audio[:len(strechedaudio)]
path = os.path.join(os.getcwd(), 'processed', vid.id + '- click.wav')
path = os.path.join(os.getcwd(), 'processed', sanitize_filename(vid.id) + '- click.wav')
sf.write(path, clicks_audio, sr)

if extractMidi:
Expand Down