diff --git a/.gitignore b/.gitignore
index b1db3f291983c114d523b100e9c3ae149e731c23..08ab719c7ab57a2c8438443652cd11f1572c5f0e 100644
--- a/.gitignore
+++ b/.gitignore
@@ -3,4 +3,5 @@ data/
env/
media/
models/
-
+build/
+*.egg-info
\ No newline at end of file
diff --git a/MANIFEST.in b/MANIFEST.in
new file mode 100644
index 0000000000000000000000000000000000000000..7c6af3d33e05eaea206a2d73856b5224bcbe2b3d
--- /dev/null
+++ b/MANIFEST.in
@@ -0,0 +1,4 @@
+include requirements.txt
+include README.md
+include preprocess_media.sh
+recursive-include autokara *
diff --git a/autokara/__init__.py b/autokara/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..b31ecc406afe6e17bb27fe80f20ed10e8516b16f
--- /dev/null
+++ b/autokara/__init__.py
@@ -0,0 +1,2 @@
+
+__version__ = "0.1.0"
\ No newline at end of file
diff --git a/autokara/autokara.py b/autokara/autokara.py
index 8fe6f2e6eafc35e54f1f1c6dab83e3f1bab72dce..e5d9e21da7dc6c38ac0a331baa7ecdbb968fe7db 100644
--- a/autokara/autokara.py
+++ b/autokara/autokara.py
@@ -5,64 +5,90 @@ import subprocess
import shlex
from pathlib import Path
-from autosyl.assUtils import AssWriter, getSyls, getHeader
-from autosyl.segment import segment
+from .autosyl.assUtils import AssWriter, getSyls, getHeader
+from .autosyl.segment import segment
-parser = argparse.ArgumentParser(description='AutoKara - Automatic karaoke timing tool')
-parser.add_argument("source_file", type=str, help="The video/audio file to time")
-parser.add_argument("ass_file", type=str, help="The ASS file with lyrics to time")
-parser.add_argument("--vocals", action="store_true", help="Treat the input as vocals file, i.e. do not perform vocals extraction")
-parser.add_argument("-o", "--output", help="Write output to specified file. If absent, overwrite source file")
-parser.add_argument("-v","--verbose", action="store_true", help="Increased verbosity")
-parser.add_argument("-l","--lang", help="Select language to use (default is Japanese Romaji)")
-args = parser.parse_args()
+def main(opts=None):
-ass_file = args.ass_file
-verbose = args.verbose
+ from g2p import __path__ as g2p_path
-if not args.vocals :
- print("Extracting audio from video file...")
- Path("./media/audio").mkdir(parents=True, exist_ok=True)
- basename = Path(args.source_file).stem
- audio_file = "media/audio/%s.wav" % basename
+ HERE = Path(__file__).parent
+ g2p_base = Path(g2p_path[0])
+ if not Path.exists(g2p_base / "mappings/langs/rji"):
+ print("No Romaji language mapping found, attempting first-time setup")
+ mapping_dir = HERE / "g2p/mappings/langs/"
+ mappings = glob.glob(f"{str(mapping_dir):s}/*")
+ for map in mappings:
+ subprocess.check_call(shlex.split(f'cp -r {map:s} {str(g2p_base):s}/mappings/langs/'))
+ subprocess.check_call(shlex.split(f'g2p update'))
+
+ if not Path.exists(g2p_base / "mappings/langs/rji"):
+ print("ERROR : Failed to find language mapping")
+ else:
+ print("Setup successful")
- subprocess.call(shlex.split('./extractWav.sh "%s" "%s"' % (args.source_file, audio_file)))
- Path("./media/vocals").mkdir(parents=True, exist_ok=True)
- output_folder = "./media/vocals"
- print("Isolating vocals...")
+ parser = argparse.ArgumentParser(description='AutoKara - Automatic karaoke timing tool')
+ parser.add_argument("source_file", type=str, help="The video/audio file to time")
+ parser.add_argument("ass_file", type=str, help="The ASS file with lyrics to time")
+ parser.add_argument("--vocals", action="store_true", help="Treat the input as vocals file, i.e. do not perform vocals extraction")
+ parser.add_argument("-o", "--output", help="Write output to specified file. If absent, overwrite source file")
+ parser.add_argument("-v","--verbose", action="store_true", help="Increased verbosity")
+ parser.add_argument("-l","--lang", help="Select language to use (default is Japanese Romaji)")
- # Not working, don't know why
- # demucs.separate.main(shlex.split('--two-stems vocals -o "%s" "%s"' % (output_folder, audio_file)))
- subprocess.call(shlex.split('demucs --two-stems vocals -o "%s" "%s"' % (output_folder, audio_file)))
+ args = parser.parse_args(opts)
- vocals_file = "./media/vocals/htdemucs/%s/vocals.wav" % basename
-else:
- vocals_file = args.source_file
+ ass_file = args.ass_file
+ verbose = args.verbose
+ if not args.vocals :
+ print("Extracting audio from video file...")
+ Path("./media/audio").mkdir(parents=True, exist_ok=True)
+ basename = Path(args.source_file).stem
+ audio_file = "media/audio/%s.wav" % basename
+ subprocess.call(shlex.split('./extractWav.sh "%s" "%s"' % (args.source_file, audio_file)))
-print("Identifying syl starts...")
+ Path("./media/vocals").mkdir(parents=True, exist_ok=True)
+ output_folder = "./media/vocals"
+ print("Isolating vocals...")
-if verbose:
- print("Retrieving syls from lyrics...")
-reference_syls, line_meta = getSyls(ass_file)
+ # Not working, don't know why
+ # demucs.separate.main(shlex.split('--two-stems vocals -o "%s" "%s"' % (output_folder, audio_file)))
+ subprocess.call(shlex.split('demucs --two-stems vocals -o "%s" "%s"' % (output_folder, audio_file)))
-if verbose:
- print("Starting syl detection...")
-syls = segment(vocals_file, reference_syls=reference_syls, verbose=verbose, language=args.lang)
-print(syls)
-print(line_meta)
+ vocals_file = "./media/vocals/htdemucs/%s/vocals.wav" % basename
+ else:
+ vocals_file = args.source_file
-print("Syls found, writing ASS file...")
-header = getHeader(ass_file)
-writer = AssWriter()
-writer.openAss(args.output if args.output else ass_file)
-writer.writeHeader(header=header)
-writer.writeSyls(syls, line_meta)
-writer.closeAss()
+
+
+ print("Identifying syl starts...")
+
+
+ if verbose:
+ print("Retrieving syls from lyrics...")
+ reference_syls, line_meta = getSyls(ass_file)
+
+ if verbose:
+ print("Starting syl detection...")
+ syls = segment(vocals_file, reference_syls=reference_syls, verbose=verbose, language=args.lang)
+ print(syls)
+ print(line_meta)
+
+ print("Syls found, writing ASS file...")
+ header = getHeader(ass_file)
+ writer = AssWriter()
+ writer.openAss(args.output if args.output else ass_file)
+ writer.writeHeader(header=header)
+ writer.writeSyls(syls, line_meta)
+ writer.closeAss()
+
+
+if __name__ == "__main__":
+ main()
diff --git a/autokara/autosyl/LyricsAlignment/model.py b/autokara/autosyl/LyricsAlignment/model.py
index f6fd66b10cb934ee1911b50d35b6bbcd13a9320f..50120c275151ad17e553ee01eb53a9d76c420962 100644
--- a/autokara/autosyl/LyricsAlignment/model.py
+++ b/autokara/autosyl/LyricsAlignment/model.py
@@ -4,7 +4,7 @@ import torch.nn.functional as F
import torchaudio
import warnings
-from autosyl.LyricsAlignment.utils import notes_to_pc
+from .utils import notes_to_pc
# following FFT parameters are designed for a 22.5k sampling rate
sr = 22050
diff --git a/autokara/autosyl/LyricsAlignment/wrapper.py b/autokara/autosyl/LyricsAlignment/wrapper.py
index 9e43b4bbe9c59074141669cb7aa11f9e357f91a6..f308ef9872955f4fdf4d45ce384c814fbee8664d 100644
--- a/autokara/autosyl/LyricsAlignment/wrapper.py
+++ b/autokara/autosyl/LyricsAlignment/wrapper.py
@@ -5,8 +5,8 @@ import torch
import torch.nn as nn
import torch.nn.functional as F
-import autosyl.LyricsAlignment.utils as utils
-from autosyl.LyricsAlignment.model import train_audio_transforms, AcousticModel, BoundaryDetection
+from . import utils
+from .model import train_audio_transforms, AcousticModel, BoundaryDetection
np.random.seed(7)
diff --git a/autokara/autosyl/segment.py b/autokara/autosyl/segment.py
index 6a54e36c971c2c60596bae459872a58418b63000..84a5ac9bb42f4e6e2354d747cc9c80402255eff2 100644
--- a/autokara/autosyl/segment.py
+++ b/autokara/autosyl/segment.py
@@ -5,9 +5,10 @@ import re
import matplotlib.pyplot as plt
import scipy.signal as sg
import parselmouth
+from pathlib import Path
-from autosyl.assUtils import getSyls, timeToDate, dateToTime
-from autosyl.LyricsAlignment.wrapper import align, preprocess_from_file
+from .assUtils import getSyls, timeToDate, dateToTime
+from .LyricsAlignment.wrapper import align, preprocess_from_file
@@ -22,7 +23,7 @@ def segment(songfile, reference_syls=None, syls_per_line=10, last_syl_dur=500, v
word_file = None # example: jamendolyrics/lyrics/*.words.txt"; Set to None if you don't have it
method = "MTL_BDR" # "Baseline", "MTL", "Baseline_BDR", "MTL_BDR"
cuda=False # set True if you have access to a GPU
- checkpoint_folder = "./autosyl/LyricsAlignment/checkpoints"
+ checkpoint_folder = f"{str(Path(__file__).parent):s}/LyricsAlignment/checkpoints"
language = language
diff --git a/autokara/plot_syls.py b/autokara/plot_syls.py
index e86960430761fc37dcc54ebfc60dbcc1a1361da1..9383639fc6cf3ab2a285eeaf9f8518ae8c53f9b5 100644
--- a/autokara/plot_syls.py
+++ b/autokara/plot_syls.py
@@ -7,8 +7,8 @@ import scipy.signal as sg
import parselmouth
import argparse
-from autosyl.assUtils import getSyls, timeToDate, dateToTime
-from autosyl.LyricsAlignment.wrapper import align, preprocess_from_file
+from .autosyl.assUtils import getSyls, timeToDate, dateToTime
+from .autosyl.LyricsAlignment.wrapper import align, preprocess_from_file
##############################################################################
@@ -20,172 +20,176 @@ from autosyl.LyricsAlignment.wrapper import align, preprocess_from_file
#
##############################################################################
+def main(opts=None):
+ parser = argparse.ArgumentParser(description='AutoKara - Automatic karaoke timing tool')
+ parser.add_argument("vocals_file", type=str, help="The audio file to time")
+ parser.add_argument("ass_file", type=str, help="The ASS file with lyrics to time")
-parser = argparse.ArgumentParser(description='AutoKara - Automatic karaoke timing tool')
-parser.add_argument("vocals_file", type=str, help="The audio file to time")
-parser.add_argument("ass_file", type=str, help="The ASS file with lyrics to time")
+ args = parser.parse_args()
-args = parser.parse_args()
+ songfile = args.vocals_file
+ reference_syls, line_meta = getSyls(sys.argv[2])
-songfile = args.vocals_file
-reference_syls, line_meta = getSyls(sys.argv[2])
+ print(reference_syls)
-print(reference_syls)
+ backtrack = False
-backtrack = False
+ audio_file = songfile # pre-computed source-separated vocals; These models do not work with mixture input.
+ word_file = None # example: jamendolyrics/lyrics/*.words.txt"; Set to None if you don't have it
+ method = "MTL_BDR" # "Baseline", "MTL", "Baseline_BDR", "MTL_BDR"
+ cuda=True # set True if you have access to a GPU
+ checkpoint_folder = "./autosyl/LyricsAlignment/checkpoints"
-audio_file = songfile # pre-computed source-separated vocals; These models do not work with mixture input.
-word_file = None # example: jamendolyrics/lyrics/*.words.txt"; Set to None if you don't have it
-method = "MTL_BDR" # "Baseline", "MTL", "Baseline_BDR", "MTL_BDR"
-cuda=True # set True if you have access to a GPU
-checkpoint_folder = "./autosyl/LyricsAlignment/checkpoints"
+ pred_file = "./MTL.csv" # saved alignment results, "(float) start_time, (float) end_time, (string) word"
-pred_file = "./MTL.csv" # saved alignment results, "(float) start_time, (float) end_time, (string) word"
+ lyrics_lines = [" ".join([syl[1] for syl in line]) for line in reference_syls]
+ #print(lyrics_lines)
-lyrics_lines = [" ".join([syl[1] for syl in line]) for line in reference_syls]
-#print(lyrics_lines)
+ # load audio and lyrics
+ # words: a list of words
+ # lyrics_p: phoneme sequence of the target lyrics
+ # idx_word_p: indices of word start in lyrics_p
+ # idx_line_p: indices of line start in lyrics_p
+ audio, words, lyrics_p, idx_word_p, idx_line_p = preprocess_from_file(audio_file, lyrics_lines, word_file)
-# load audio and lyrics
-# words: a list of words
-# lyrics_p: phoneme sequence of the target lyrics
-# idx_word_p: indices of word start in lyrics_p
-# idx_line_p: indices of line start in lyrics_p
-audio, words, lyrics_p, idx_word_p, idx_line_p = preprocess_from_file(audio_file, lyrics_lines, word_file)
+ # compute alignment
+ # word_align: a list of frame indices aligned to each word
+ # words: a list of words
+ word_align, words = align(audio, words, lyrics_p, idx_word_p, idx_line_p, method=method, cuda=False, checkpoint_folder=checkpoint_folder)
-# compute alignment
-# word_align: a list of frame indices aligned to each word
-# words: a list of words
-word_align, words = align(audio, words, lyrics_p, idx_word_p, idx_line_p, method=method, cuda=False, checkpoint_folder=checkpoint_folder)
+ print([[word_align[i][0], word_align[i][1], words[i]] for i in range(len(word_align))])
+ words_onsets = np.array([word_align[i][0] for i in range(len(word_align))])
-print([[word_align[i][0], word_align[i][1], words[i]] for i in range(len(word_align))])
-words_onsets = np.array([word_align[i][0] for i in range(len(word_align))])
+ cnn = madmom.features.onsets.CNNOnsetProcessor()
+ spectral = madmom.features.onsets.SpectralOnsetProcessor('modified_kullback_leibler')
-cnn = madmom.features.onsets.CNNOnsetProcessor()
-spectral = madmom.features.onsets.SpectralOnsetProcessor('modified_kullback_leibler')
+ sig = madmom.audio.signal.Signal(songfile, num_channels=1)
+ parsel = parselmouth.Sound(sig)
-sig = madmom.audio.signal.Signal(songfile, num_channels=1)
-parsel = parselmouth.Sound(sig)
+ spec = madmom.audio.spectrogram.Spectrogram(sig)
+ filt_spec = madmom.audio.spectrogram.FilteredSpectrogram(spec, filterbank=madmom.audio.filters.LogFilterbank, num_bands=24)
+ log_spec = madmom.audio.spectrogram.LogarithmicSpectrogram(filt_spec, add=1)
-spec = madmom.audio.spectrogram.Spectrogram(sig)
-filt_spec = madmom.audio.spectrogram.FilteredSpectrogram(spec, filterbank=madmom.audio.filters.LogFilterbank, num_bands=24)
-log_spec = madmom.audio.spectrogram.LogarithmicSpectrogram(filt_spec, add=1)
+ magnitude = np.max(log_spec[:,:100], axis=1)
-magnitude = np.max(log_spec[:,:100], axis=1)
+ cnn_function = cnn(sig)
+ spectral_function = spectral(sig)
+ spectral_function = spectral_function/(spectral_function.max())
-cnn_function = cnn(sig)
-spectral_function = spectral(sig)
-spectral_function = spectral_function/(spectral_function.max())
+ #activation_function = 0.5*cnn_function + 0.5*spectral_function
+ activation_function = (2 * cnn_function * spectral_function)/(cnn_function + spectral_function)
+ #activation_function = np.where(spectral_function > 0.14, cnn_function, 0)
+ #onsets = proc(activation_function)
-#activation_function = 0.5*cnn_function + 0.5*spectral_function
-activation_function = (2 * cnn_function * spectral_function)/(cnn_function + spectral_function)
-#activation_function = np.where(spectral_function > 0.14, cnn_function, 0)
-#onsets = proc(activation_function)
+ if reference_syls:
+ activation_threshold = 0.1
+ else:
+ activation_threshold = 0.2
-if reference_syls:
- activation_threshold = 0.1
-else:
- activation_threshold = 0.2
-
-activation_smoothed = madmom.audio.signal.smooth(activation_function, 20)
-cnn_smoothed = madmom.audio.signal.smooth(cnn_function, 20)
-onsets = madmom.features.onsets.peak_picking(activation_smoothed, threshold=activation_threshold, smooth=0)
-#onsets = np.array([o for o in onsets if cnn_smoothed[o] > 0.1])
-
-pitch = parsel.to_pitch()
-pitch_values = pitch.selected_array['frequency']
-
-pad_before = round(pitch.xs()[0]*100)
-pad_after = len(magnitude) - len(pitch_values) - pad_before
-
-pitch_values = np.pad(pitch_values, (pad_before, pad_after), 'constant', constant_values=(0,0))
-
-mask_function = magnitude * pitch_values
-mask_function = mask_function/np.max(mask_function)
-mask_threshold = 0.15
-mask_window = [1,6]
-invalid_onsets_idx = []
-
-for i in range(len(onsets)):
- if np.max(mask_function[onsets[i]+mask_window[0]:onsets[i]+mask_window[1]]) < mask_threshold:
- invalid_onsets_idx.append(i)
-
-onsets = np.delete(onsets, invalid_onsets_idx)
+ activation_smoothed = madmom.audio.signal.smooth(activation_function, 20)
+ cnn_smoothed = madmom.audio.signal.smooth(cnn_function, 20)
+ onsets = madmom.features.onsets.peak_picking(activation_smoothed, threshold=activation_threshold, smooth=0)
+ #onsets = np.array([o for o in onsets if cnn_smoothed[o] > 0.1])
+ pitch = parsel.to_pitch()
+ pitch_values = pitch.selected_array['frequency']
+ pad_before = round(pitch.xs()[0]*100)
+ pad_after = len(magnitude) - len(pitch_values) - pad_before
-if reference_syls:
- filtered_onsets = []
- line_index = 0
- for line in reference_syls:
- line_index += 1
- syl_number = len(line) - 1
- line_onsets = [o for o in onsets if (o >= line[0][0] and o <= line[-1][0])]
- line_onsets.sort(reverse=True, key=(lambda x: activation_smoothed[x]))
- if syl_number > len(line_onsets):
- print("WARNING : failed to detect enough onsets in line %d (%d, %d)" % (line_index, line[0][0], line[-1][0]))
- filtered_onsets += line_onsets[0:syl_number]
-
- onsets = np.array(sorted(filtered_onsets))
+ pitch_values = np.pad(pitch_values, (pad_before, pad_after), 'constant', constant_values=(0,0))
+ mask_function = magnitude * pitch_values
+ mask_function = mask_function/np.max(mask_function)
+ mask_threshold = 0.15
+ mask_window = [1,6]
+ invalid_onsets_idx = []
-"""
- if word_index > 0:
- word_start = max(word_align[word_index][0] - 5, line[0][0], previous_onset+1)
- else:
- word_start = line[0][0]
- if word_index < len(words) - 1 and syl_index < len(line) - 2:
- word_end = min(line[-1][0], word_align[word_index + 1][0] - 5)
- else:
- word_end = line[-1][0]
-
- word_onsets = [o for o in onsets if (o >= word_start and o <= word_end)]
- word_onsets.sort(reverse=True, key=(lambda x: activation_smoothed[x]))
- if word_syl_count > len(word_onsets):
- print("WARNING : failed to detect enough onsets in word %s (%d, %d)" % (word_tmp, word_start, word_end))
- filtered_onsets += word_onsets[0:word_syl_count]
- print(word_onsets[0:word_syl_count])
- previous_onset = max(word_onsets[0:word_syl_count] + [0])
-"""
-
-# Backtrack onsets to closest earlier local minimum
-if backtrack:
- backtrack_max_frames = 50
for i in range(len(onsets)):
- initial_onset = onsets[i]
- while(activation_smoothed[onsets[i] - 1] < activation_smoothed[onsets[i]] and onsets[i] > initial_onset - backtrack_max_frames):
- onsets[i] -= 1
-
-#print(onsets/100)
-print(words_onsets/100)
-
-if reference_syls:
- reference_onsets = [syl[0]+8 for line in reference_syls for syl in line[:-1]]
-
-fig, axs = plt.subplots(nrows=2, sharex=True)
-axs[0].imshow(log_spec.T, origin='lower', aspect='auto')
-if reference_syls:
- axs[0].vlines(reference_onsets, 0, 140, colors='red')
-axs[0].plot((pitch_values/np.max(pitch_values))*140, color='yellow')
-axs[1].plot(mask_function)
-#axs[1].plot(cnn_smoothed)
-#axs[1].plot(spectral_function, color='green')
-axs[1].plot(activation_smoothed, color='orange')
-axs[1].vlines(onsets, 0, 2, colors='red')
-axs[1].vlines(words_onsets, 0, 3, colors='m')
-axs[1].hlines([max(mask_threshold, 0), activation_threshold], 0, onsets[-1]+100, colors='black')
-
-#bins = np.arange(0, 1, 0.02)
-#hist, hist_axs = plt.subplots(nrows=1)
-#hist_axs.hist(mask_function, bins=bins)
-
-plt.show()
\ No newline at end of file
+ if np.max(mask_function[onsets[i]+mask_window[0]:onsets[i]+mask_window[1]]) < mask_threshold:
+ invalid_onsets_idx.append(i)
+
+ onsets = np.delete(onsets, invalid_onsets_idx)
+
+
+
+ if reference_syls:
+ filtered_onsets = []
+ line_index = 0
+ for line in reference_syls:
+ line_index += 1
+ syl_number = len(line) - 1
+ line_onsets = [o for o in onsets if (o >= line[0][0] and o <= line[-1][0])]
+ line_onsets.sort(reverse=True, key=(lambda x: activation_smoothed[x]))
+ if syl_number > len(line_onsets):
+ print("WARNING : failed to detect enough onsets in line %d (%d, %d)" % (line_index, line[0][0], line[-1][0]))
+ filtered_onsets += line_onsets[0:syl_number]
+
+ onsets = np.array(sorted(filtered_onsets))
+
+
+ """
+ if word_index > 0:
+ word_start = max(word_align[word_index][0] - 5, line[0][0], previous_onset+1)
+ else:
+ word_start = line[0][0]
+ if word_index < len(words) - 1 and syl_index < len(line) - 2:
+ word_end = min(line[-1][0], word_align[word_index + 1][0] - 5)
+ else:
+ word_end = line[-1][0]
+
+ word_onsets = [o for o in onsets if (o >= word_start and o <= word_end)]
+ word_onsets.sort(reverse=True, key=(lambda x: activation_smoothed[x]))
+ if word_syl_count > len(word_onsets):
+ print("WARNING : failed to detect enough onsets in word %s (%d, %d)" % (word_tmp, word_start, word_end))
+ filtered_onsets += word_onsets[0:word_syl_count]
+ print(word_onsets[0:word_syl_count])
+ previous_onset = max(word_onsets[0:word_syl_count] + [0])
+ """
+
+ # Backtrack onsets to closest earlier local minimum
+ if backtrack:
+ backtrack_max_frames = 50
+ for i in range(len(onsets)):
+ initial_onset = onsets[i]
+ while(activation_smoothed[onsets[i] - 1] < activation_smoothed[onsets[i]] and onsets[i] > initial_onset - backtrack_max_frames):
+ onsets[i] -= 1
+
+ #print(onsets/100)
+ print(words_onsets/100)
+
+ if reference_syls:
+ reference_onsets = [syl[0]+8 for line in reference_syls for syl in line[:-1]]
+
+ fig, axs = plt.subplots(nrows=2, sharex=True)
+ axs[0].imshow(log_spec.T, origin='lower', aspect='auto')
+ if reference_syls:
+ axs[0].vlines(reference_onsets, 0, 140, colors='red')
+ axs[0].plot((pitch_values/np.max(pitch_values))*140, color='yellow')
+ axs[1].plot(mask_function)
+ #axs[1].plot(cnn_smoothed)
+ #axs[1].plot(spectral_function, color='green')
+ axs[1].plot(activation_smoothed, color='orange')
+ axs[1].vlines(onsets, 0, 2, colors='red')
+ axs[1].vlines(words_onsets, 0, 3, colors='m')
+ axs[1].hlines([max(mask_threshold, 0), activation_threshold], 0, onsets[-1]+100, colors='black')
+
+ #bins = np.arange(0, 1, 0.02)
+ #hist, hist_axs = plt.subplots(nrows=1)
+ #hist_axs.hist(mask_function, bins=bins)
+
+ plt.show()
+
+
+if __name__ == "__main__":
+ main()
\ No newline at end of file
diff --git a/autokara/update_lang_db.py b/autokara/update_lang_db.py
new file mode 100644
index 0000000000000000000000000000000000000000..6e0c63e926ebef29287967a04d7fbed6c53cb9f8
--- /dev/null
+++ b/autokara/update_lang_db.py
@@ -0,0 +1,28 @@
+import subprocess
+import shlex
+from pathlib import Path
+import glob
+
+
+def main():
+ from g2p import __path__ as g2p_path
+
+ HERE = Path(__file__).parent
+ g2p_base = Path(g2p_path[0])
+
+ print("Regenerating custom language mappings...")
+ mapping_dir = HERE / "g2p/mappings/langs/"
+ mappings = glob.glob(f"{str(mapping_dir):s}/*")
+ for map in mappings:
+ subprocess.check_call(shlex.split(f'cp -r {map:s} {str(g2p_base):s}/mappings/langs/'))
+ subprocess.check_call(shlex.split(f'g2p update'))
+
+ if not Path.exists(g2p_base / "mappings/langs/rji"):
+ print("ERROR : Failed to find language mapping")
+ else:
+ print("Setup successful")
+
+
+
+if __name__ == "__main__":
+ main()
diff --git a/requirements.txt b/requirements.txt
index 04299146d72785e9cf25e747f7bcea65dd11d05c..1574a1b402f57d64880c618eb05472c3fd8879fc 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -1,15 +1,9 @@
librosa
demucs
-chainer
soundfile
-sklearn
matplotlib
numpy
-tqdm
-scipy
-cython
-mido
-git+https://github.com/CPJKU/madmom.git
+madmom@git+https://github.com/CPJKU/madmom.git
praat-parselmouth
future
musdb
diff --git a/setup.py b/setup.py
new file mode 100644
index 0000000000000000000000000000000000000000..6f9b33985caa38b4b5217591a26ac3df162efb1c
--- /dev/null
+++ b/setup.py
@@ -0,0 +1,64 @@
+
+from pathlib import Path
+from setuptools import setup, find_packages
+import atexit
+from setuptools.command.install import install
+import subprocess
+import shlex
+import glob
+
+
+NAME = 'autokara'
+DESCRIPTION = 'Automatic karaoke timing'
+
+URL = 'https://git.iiens.net/bakaclub/autokara'
+AUTHOR = 'Loïc "Sting" Allègre'
+REQUIRES_PYTHON = '>=3.8.0'
+
+HERE = Path(__file__).parent
+
+# Get version without explicitly loading the module.
+for line in open('autokara/__init__.py'):
+ line = line.strip()
+ if '__version__' in line:
+ context = {}
+ exec(line, context)
+ VERSION = context['__version__']
+
+
+def load_requirements(name):
+ required = [i.strip() for i in open(HERE / name)]
+ required = [i for i in required if not i.startswith('#')]
+ print(required)
+ return required
+
+
+REQUIRED = load_requirements('requirements.txt')
+ALL_REQUIRED = load_requirements('requirements.txt')
+
+
+setup(
+ name=NAME,
+ version=VERSION,
+ description=DESCRIPTION,
+ author=AUTHOR,
+ python_requires=REQUIRES_PYTHON,
+ url=URL,
+ packages=find_packages(),
+ install_requires=REQUIRED,
+ include_package_data=True,
+ entry_points={
+ 'console_scripts': ['autokara=autokara.autokara:main',
+ 'autokara-plot=autokara.plot_syls:main',
+ 'autokara-gen-lang=autokara.update_lang_db:main'
+ ],
+ },
+ license='MIT License',
+ classifiers=[
+ # Trove classifiers
+ # Full list: https://pypi.python.org/pypi?%3Aaction=list_classifiers
+ 'License :: OSI Approved :: MIT License',
+ 'Topic :: Multimedia :: Sound/Audio',
+ 'Topic :: Scientific/Engineering :: Artificial Intelligence',
+ ],
+)