-init-first-stage-
This commit is contained in:
parent
55de001c95
commit
0a297e331d
20
docs/document.aux
Normal file
20
docs/document.aux
Normal file
@ -0,0 +1,20 @@
|
||||
\relax
|
||||
\@writefile{toc}{\contentsline {section}{\numberline {1}Wst\IeC {\k e}p}{2}\protected@file@percent }
|
||||
\@writefile{toc}{\contentsline {subsection}{\numberline {1.1}Muzyka}{2}\protected@file@percent }
|
||||
\@writefile{toc}{\contentsline {section}{\numberline {2}MIDI, Muzyka jako Informacje}{2}\protected@file@percent }
|
||||
\@writefile{toc}{\contentsline {subsection}{\numberline {2.1}MIDI}{2}\protected@file@percent }
|
||||
\@writefile{toc}{\contentsline {subsection}{\numberline {2.2}Pianoroll}{2}\protected@file@percent }
|
||||
\@writefile{toc}{\contentsline {subsection}{\numberline {2.3}Muzyka jako tr\IeC {\'o}jwymiarowa tablica}{2}\protected@file@percent }
|
||||
\@writefile{toc}{\contentsline {section}{\numberline {3}Generatwne sieci neuronowe - GANy, VAE, LSTMy}{2}\protected@file@percent }
|
||||
\@writefile{toc}{\contentsline {subsection}{\numberline {3.1}Autoencodery, VAE}{2}\protected@file@percent }
|
||||
\@writefile{toc}{\contentsline {subsection}{\numberline {3.2}LSTM}{2}\protected@file@percent }
|
||||
\@writefile{toc}{\contentsline {section}{\numberline {4}Modele generatywne stosowane w generowaniu muzyki}{3}\protected@file@percent }
|
||||
\@writefile{toc}{\contentsline {subsection}{\numberline {4.1}Project Magenta}{3}\protected@file@percent }
|
||||
\@writefile{toc}{\contentsline {subsection}{\numberline {4.2}MuseGAN}{3}\protected@file@percent }
|
||||
\@writefile{toc}{\contentsline {subsection}{\numberline {4.3}VAE-MIDI}{3}\protected@file@percent }
|
||||
\@writefile{toc}{\contentsline {section}{\numberline {5}Budowanie generatora muzyki}{3}\protected@file@percent }
|
||||
\@writefile{toc}{\contentsline {subsection}{\numberline {5.1}Wyodr\IeC {\k e}bnienie danych z plik\IeC {\'o}w MIDI}{3}\protected@file@percent }
|
||||
\@writefile{toc}{\contentsline {subsection}{\numberline {5.2}Przygotowanie Modelu GAN}{3}\protected@file@percent }
|
||||
\@writefile{toc}{\contentsline {subsection}{\numberline {5.3}Proces uczenia, pr\IeC {\'o}bki co kilka epoch\IeC {\'o}w, costloss wykres}{3}\protected@file@percent }
|
||||
\@writefile{toc}{\contentsline {subsection}{\numberline {5.4}Pr\IeC {\'o}bki ko\IeC {\'n}cowe, jak\IeC {\k a} muzyk\IeC {\k e} da si\IeC {\k e} z tego wygenerowa\IeC {\'c}}{3}\protected@file@percent }
|
||||
\@writefile{toc}{\contentsline {section}{\numberline {6}Podsumowanie}{3}\protected@file@percent }
|
||||
151
docs/document.log
Normal file
151
docs/document.log
Normal file
@ -0,0 +1,151 @@
|
||||
This is pdfTeX, Version 3.14159265-2.6-1.40.19 (TeX Live 2018/W32TeX) (preloaded format=pdflatex 2019.2.21) 28 MAY 2019 12:32
|
||||
entering extended mode
|
||||
restricted \write18 enabled.
|
||||
%&-line parsing enabled.
|
||||
**document.tex
|
||||
(./document.tex
|
||||
LaTeX2e <2018-12-01>
|
||||
(c:/software/latex/texmf-dist/tex/latex/base/article.cls
|
||||
Document Class: article 2018/09/03 v1.4i Standard LaTeX document class
|
||||
(c:/software/latex/texmf-dist/tex/latex/base/size10.clo
|
||||
File: size10.clo 2018/09/03 v1.4i Standard LaTeX file (size option)
|
||||
)
|
||||
\c@part=\count80
|
||||
\c@section=\count81
|
||||
\c@subsection=\count82
|
||||
\c@subsubsection=\count83
|
||||
\c@paragraph=\count84
|
||||
\c@subparagraph=\count85
|
||||
\c@figure=\count86
|
||||
\c@table=\count87
|
||||
\abovecaptionskip=\skip41
|
||||
\belowcaptionskip=\skip42
|
||||
\bibindent=\dimen102
|
||||
)
|
||||
(c:/software/latex/texmf-dist/tex/latex/polski/polski.sty
|
||||
Package: polski 2017/05/04 v1.3.4 Polish language package
|
||||
|
||||
Switching to Polish text encoding and Polish maths fonts.
|
||||
(c:/software/latex/texmf-dist/tex/latex/base/ot4enc.def
|
||||
File: ot4enc.def 2018/08/11 v2.0j Standard LaTeX file
|
||||
Now handling font encoding OT4 ...
|
||||
... no UTF-8 mapping file for font encoding OT4
|
||||
)
|
||||
LaTeX Font Info: Try loading font information for OT4+cmr on input line 360.
|
||||
|
||||
|
||||
(c:/software/latex/texmf-dist/tex/latex/polski/ot4cmr.fd
|
||||
File: ot4cmr.fd 2008/02/24 v1.2.1 Font defs for fonts PL (MW)
|
||||
)
|
||||
LaTeX Font Info: Overwriting math alphabet `\mathbf' in version `normal'
|
||||
(Font) OT1/cmr/bx/n --> OT4/cmr/bx/n on input line 360.
|
||||
LaTeX Font Info: Overwriting math alphabet `\mathsf' in version `normal'
|
||||
(Font) OT1/cmss/m/n --> OT4/cmss/m/n on input line 360.
|
||||
LaTeX Font Info: Overwriting math alphabet `\mathit' in version `normal'
|
||||
(Font) OT1/cmr/m/it --> OT4/cmr/m/it on input line 360.
|
||||
LaTeX Font Info: Overwriting math alphabet `\mathtt' in version `normal'
|
||||
(Font) OT1/cmtt/m/n --> OT4/cmtt/m/n on input line 360.
|
||||
LaTeX Font Info: Overwriting math alphabet `\mathsf' in version `bold'
|
||||
(Font) OT1/cmss/bx/n --> OT4/cmss/bx/n on input line 360.
|
||||
LaTeX Font Info: Overwriting math alphabet `\mathit' in version `bold'
|
||||
(Font) OT1/cmr/bx/it --> OT4/cmr/bx/it on input line 360.
|
||||
LaTeX Font Info: Encoding `OT1' has changed to `OT4' for symbol font
|
||||
(Font) `operators' in the math version `normal' on input line 360.
|
||||
|
||||
LaTeX Font Info: Overwriting symbol font `operators' in version `normal'
|
||||
(Font) OT1/cmr/m/n --> OT4/cmr/m/n on input line 360.
|
||||
LaTeX Font Info: Overwriting symbol font `letters' in version `normal'
|
||||
(Font) OML/cmm/m/it --> OML/plm/m/it on input line 360.
|
||||
LaTeX Font Info: Overwriting symbol font `symbols' in version `normal'
|
||||
(Font) OMS/cmsy/m/n --> OMS/plsy/m/n on input line 360.
|
||||
LaTeX Font Info: Overwriting symbol font `largesymbols' in version `normal'
|
||||
(Font) OMX/cmex/m/n --> OMX/plex/m/n on input line 360.
|
||||
LaTeX Font Info: Encoding `OT1' has changed to `OT4' for symbol font
|
||||
(Font) `operators' in the math version `bold' on input line 360.
|
||||
LaTeX Font Info: Overwriting symbol font `operators' in version `bold'
|
||||
(Font) OT1/cmr/bx/n --> OT4/cmr/bx/n on input line 360.
|
||||
LaTeX Font Info: Overwriting symbol font `letters' in version `bold'
|
||||
(Font) OML/cmm/b/it --> OML/plm/b/it on input line 360.
|
||||
LaTeX Font Info: Overwriting symbol font `symbols' in version `bold'
|
||||
(Font) OMS/cmsy/b/n --> OMS/plsy/b/n on input line 360.
|
||||
)
|
||||
|
||||
LaTeX Warning: Unused global option(s):
|
||||
[utf8].
|
||||
|
||||
(./document.aux)
|
||||
\openout1 = `document.aux'.
|
||||
|
||||
LaTeX Font Info: Checking defaults for OML/cmm/m/it on input line 18.
|
||||
LaTeX Font Info: ... okay on input line 18.
|
||||
LaTeX Font Info: Checking defaults for T1/cmr/m/n on input line 18.
|
||||
LaTeX Font Info: ... okay on input line 18.
|
||||
LaTeX Font Info: Checking defaults for OT1/cmr/m/n on input line 18.
|
||||
LaTeX Font Info: ... okay on input line 18.
|
||||
LaTeX Font Info: Checking defaults for OMS/cmsy/m/n on input line 18.
|
||||
LaTeX Font Info: ... okay on input line 18.
|
||||
LaTeX Font Info: Checking defaults for OMX/cmex/m/n on input line 18.
|
||||
LaTeX Font Info: ... okay on input line 18.
|
||||
LaTeX Font Info: Checking defaults for U/cmr/m/n on input line 18.
|
||||
LaTeX Font Info: ... okay on input line 18.
|
||||
LaTeX Font Info: Checking defaults for OT4/cmr/m/n on input line 18.
|
||||
LaTeX Font Info: ... okay on input line 18.
|
||||
LaTeX Font Info: Try loading font information for OML+plm on input line 20.
|
||||
(c:/software/latex/texmf-dist/tex/latex/polski/omlplm.fd
|
||||
File: omlplm.fd 2008/02/24 v1.2.1 Font defs for fonts PL (MW)
|
||||
)
|
||||
LaTeX Font Info: Try loading font information for OMS+plsy on input line 20.
|
||||
|
||||
|
||||
(c:/software/latex/texmf-dist/tex/latex/polski/omsplsy.fd
|
||||
File: omsplsy.fd 2008/02/24 v1.2.1 Font defs for fonts PL (MW)
|
||||
)
|
||||
LaTeX Font Info: Try loading font information for OMX+plex on input line 20.
|
||||
|
||||
|
||||
(c:/software/latex/texmf-dist/tex/latex/polski/omxplex.fd
|
||||
File: omxplex.fd 2008/02/24 v1.2.1 Font defs for fonts PL (MW)
|
||||
)
|
||||
LaTeX Font Info: External font `plex10' loaded for size
|
||||
(Font) <12> on input line 20.
|
||||
LaTeX Font Info: External font `plex10' loaded for size
|
||||
(Font) <8> on input line 20.
|
||||
LaTeX Font Info: External font `plex10' loaded for size
|
||||
(Font) <6> on input line 20.
|
||||
[1
|
||||
|
||||
{c:/software/latex/texmf-var/fonts/map/pdftex/updmap/pdftex.map}] (./document.t
|
||||
oc
|
||||
LaTeX Font Info: External font `plex10' loaded for size
|
||||
(Font) <10> on input line 2.
|
||||
LaTeX Font Info: External font `plex10' loaded for size
|
||||
(Font) <7> on input line 2.
|
||||
LaTeX Font Info: External font `plex10' loaded for size
|
||||
(Font) <5> on input line 2.
|
||||
)
|
||||
\tf@toc=\write3
|
||||
\openout3 = `document.toc'.
|
||||
|
||||
[1] [2]
|
||||
[3] (./document.aux) )
|
||||
Here is how much of TeX's memory you used:
|
||||
474 strings out of 492616
|
||||
5750 string characters out of 6131816
|
||||
66703 words of memory out of 5000000
|
||||
4428 multiletter control sequences out of 15000+600000
|
||||
13779 words of font info for 39 fonts, out of 8000000 for 9000
|
||||
1141 hyphenation exceptions out of 8191
|
||||
23i,7n,25p,379b,252s stack positions out of 5000i,500n,10000p,200000b,80000s
|
||||
{c:/software/latex/texmf-dist/fonts/enc/dvips/pl/plrm.enc
|
||||
}<c:/software/latex/texmf-dist/fonts/type1/public/pl/plbx10.pfb><c:/software/la
|
||||
tex/texmf-dist/fonts/type1/public/pl/plbx12.pfb><c:/software/latex/texmf-dist/f
|
||||
onts/type1/public/pl/plr10.pfb><c:/software/latex/texmf-dist/fonts/type1/public
|
||||
/pl/plr12.pfb><c:/software/latex/texmf-dist/fonts/type1/public/pl/plr17.pfb><c:
|
||||
/software/latex/texmf-dist/fonts/type1/public/pl/plr9.pfb>
|
||||
Output written on document.pdf (4 pages, 61011 bytes).
|
||||
PDF statistics:
|
||||
42 PDF objects out of 1000 (max. 8388607)
|
||||
29 compressed objects within 1 object stream
|
||||
0 named destinations out of 1000 (max. 500000)
|
||||
1 words of extra memory for PDF output out of 10000 (max. 10000000)
|
||||
|
||||
BIN
docs/document.pdf
Normal file
BIN
docs/document.pdf
Normal file
Binary file not shown.
BIN
docs/document.synctex.gz
Normal file
BIN
docs/document.synctex.gz
Normal file
Binary file not shown.
80
docs/document.tex
Normal file
80
docs/document.tex
Normal file
@ -0,0 +1,80 @@
|
||||
\documentclass[utf8]{article}
|
||||
\usepackage{polski}
|
||||
|
||||
\title{%
|
||||
Generowanie muzyki \\
|
||||
przy pomocy głębokiego uczenia \\
|
||||
\large Music generation with deep learning}
|
||||
|
||||
|
||||
\author{%
|
||||
Cezary Pukownik \\
|
||||
\newline
|
||||
\small Opiekun pracy:\\
|
||||
dr hab. Tomasz Górecki}
|
||||
|
||||
\date{2019-05-28}
|
||||
|
||||
\begin{document}
|
||||
|
||||
\maketitle
|
||||
|
||||
\newpage
|
||||
\pagenumbering{arabic}
|
||||
\tableofcontents
|
||||
\newpage
|
||||
|
||||
|
||||
\section{Wstęp}
|
||||
To jest wstep do pracy magisterskiej
|
||||
|
||||
\subsection{Muzyka}
|
||||
Teraz opowiem troche o muzyce, i dlaczego trudno jest ja generowac, co o tym sądze, oraz czy sztuczna inteligencja zastapi muzyków w przyszłości.
|
||||
|
||||
\section{MIDI, Muzyka jako Informacje}
|
||||
Tutaj opiszę w jaki sposób muzyka jest zapisywana jako informacje komputerowe, protokuł midi, przedstawienie muzyki jako pianorolle.
|
||||
|
||||
\subsection{MIDI}
|
||||
Tutaj opiszę protokuł MIDI
|
||||
|
||||
\subsection{Pianoroll}
|
||||
Tutaj opisze co todsdsddsdss są pianorolle, jak je czytać i czemu służą.
|
||||
|
||||
\subsection{Muzyka jako trójwymiarowa tablica}
|
||||
Tutaj opisze dlaczego muzykę moża opisać jako trójwymiarowa tablicę.
|
||||
|
||||
\section{Generatwne sieci neuronowe - GANy, VAE, LSTMy}
|
||||
Tutaj będzie opisane, dlaczego sieci neuronowe, radzą sobie lepiej w produkowaniu muzyki niż inne modele. Oraz jakie modele są odpowidnie do pewnych zastosowań, JAZZ - LSTM, bardziej ustrukturyzowana - VAE itp.
|
||||
|
||||
\subsection{Autoencodery, VAE}
|
||||
Teraz opowiem troche o muzyce, i dlaczego trudno jest ja generowac
|
||||
|
||||
\subsection{LSTM}
|
||||
Teraz opowiem troche o muzyce, i dlaczego trudno jest ja generowac
|
||||
|
||||
\section{Modele generatywne stosowane w generowaniu muzyki}
|
||||
Przykłady gotowych podeść do generowania muzyki, oraz jakie modele zostały zastosowane. dlaczego takie itp.
|
||||
|
||||
\subsection{Project Magenta}
|
||||
Teraz opowiem troche o muzyce, i dlaczego trudno jest ja generowac
|
||||
|
||||
\subsection{MuseGAN}
|
||||
Teraz opowiem troche o muzyce, i dlaczego trudno jest ja generowac
|
||||
|
||||
\subsection{VAE-MIDI}
|
||||
Teraz opowiem troche o muzyce, i dlaczego trudno jest ja generowac
|
||||
|
||||
|
||||
\section{Budowanie generatora muzyki}
|
||||
W tym rozdzialę opiszę w jaki sposób zbudowałem swój własny geneator muzyki, jak przechodził procesz uczenia, jakie próbki udało mi się wygenrować. Opis kodu który napisałem.
|
||||
|
||||
\subsection{Wyodrębnienie danych z plików MIDI}
|
||||
\subsection{Przygotowanie Modelu GAN}
|
||||
\subsection{Proces uczenia, próbki co kilka epochów, costloss wykres}
|
||||
\subsection{Próbki końcowe, jaką muzykę da się z tego wygenerować}
|
||||
|
||||
\section{Podsumowanie}
|
||||
Ostateczne wnioski, czy muzyka generowana komputerowa da się lubić? Czy to pozytywnie wpłynie na przemysł muzyczny? Tak i nie. Może złużyć jako inspiracja dla muzyków, proces wspierający. Z drugiej strony może obnizy koszty produkowania muzyki pop, która i tak jest już bardzo powtarzalna. Czy sieci neuronowe nauczą się produkować Hity?
|
||||
|
||||
|
||||
\end{document}
|
||||
19
docs/document.toc
Normal file
19
docs/document.toc
Normal file
@ -0,0 +1,19 @@
|
||||
\contentsline {section}{\numberline {1}Wst\IeC {\k e}p}{2}%
|
||||
\contentsline {subsection}{\numberline {1.1}Muzyka}{2}%
|
||||
\contentsline {section}{\numberline {2}MIDI, Muzyka jako Informacje}{2}%
|
||||
\contentsline {subsection}{\numberline {2.1}MIDI}{2}%
|
||||
\contentsline {subsection}{\numberline {2.2}Pianoroll}{2}%
|
||||
\contentsline {subsection}{\numberline {2.3}Muzyka jako tr\IeC {\'o}jwymiarowa tablica}{2}%
|
||||
\contentsline {section}{\numberline {3}Generatwne sieci neuronowe - GANy, VAE, LSTMy}{2}%
|
||||
\contentsline {subsection}{\numberline {3.1}Autoencodery, VAE}{2}%
|
||||
\contentsline {subsection}{\numberline {3.2}LSTM}{2}%
|
||||
\contentsline {section}{\numberline {4}Modele generatywne stosowane w generowaniu muzyki}{3}%
|
||||
\contentsline {subsection}{\numberline {4.1}Project Magenta}{3}%
|
||||
\contentsline {subsection}{\numberline {4.2}MuseGAN}{3}%
|
||||
\contentsline {subsection}{\numberline {4.3}VAE-MIDI}{3}%
|
||||
\contentsline {section}{\numberline {5}Budowanie generatora muzyki}{3}%
|
||||
\contentsline {subsection}{\numberline {5.1}Wyodr\IeC {\k e}bnienie danych z plik\IeC {\'o}w MIDI}{3}%
|
||||
\contentsline {subsection}{\numberline {5.2}Przygotowanie Modelu GAN}{3}%
|
||||
\contentsline {subsection}{\numberline {5.3}Proces uczenia, pr\IeC {\'o}bki co kilka epoch\IeC {\'o}w, costloss wykres}{3}%
|
||||
\contentsline {subsection}{\numberline {5.4}Pr\IeC {\'o}bki ko\IeC {\'n}cowe, jak\IeC {\k a} muzyk\IeC {\k e} da si\IeC {\k e} z tego wygenerowa\IeC {\'c}}{3}%
|
||||
\contentsline {section}{\numberline {6}Podsumowanie}{3}%
|
||||
22
generate.py
22
generate.py
@ -1,22 +0,0 @@
|
||||
import numpy as np
|
||||
from keras.layers import Input, Dense, Conv2D
|
||||
from keras.models import Model
|
||||
GENERATED_BEAT_PATH = 'data/output/generated_bar'
|
||||
MODEL_PATH = 'data/autoencoder_model.h5'
|
||||
SAMPLES_PATH = 'data/samples.npz'
|
||||
|
||||
input = Input(shape=(1,96,128))
|
||||
encoded = Conv2D(filters = 32, kernel_size = 1)(input)
|
||||
decoded = Conv2D(filters = 128, kernel_size = 1)(encoded)
|
||||
autoencoder = Model(input, decoded)
|
||||
|
||||
# load weights into new model
|
||||
autoencoder.load_weights(MODEL_PATH)
|
||||
print("Loaded model from disk")
|
||||
|
||||
# generate_seed = np.random.rand(1,1,96,128)
|
||||
|
||||
generate_seed = np.load(SAMPLES_PATH)['arr_0'][0:]
|
||||
|
||||
generated_beat = autoencoder.predict(generate_seed)
|
||||
np.savez_compressed(GENERATED_BEAT_PATH, generated_beat)
|
||||
@ -1,76 +0,0 @@
|
||||
import settings
|
||||
import pypianoroll as roll
|
||||
import matplotlib.pyplot as plt
|
||||
import numpy as np
|
||||
import os
|
||||
from math import floor
|
||||
|
||||
MIDI_DIRECTORY = settings.midi_path
|
||||
SAMPLES_DIRECTORY = settings.samples_path
|
||||
MIDI_RESOLUTION = settings.midi_resolution
|
||||
BEAT_PER_BATCH = settings.beats_per_sample
|
||||
|
||||
samples = np.empty((0,BEAT_PER_BATCH,96,128))
|
||||
|
||||
def erase_note_lenth(pianoroll):
|
||||
if pianoroll.ndim != 2:
|
||||
raise ValueError('pianoroll should be two dimentional')
|
||||
now_block = []
|
||||
for x in pianoroll:
|
||||
this = None
|
||||
prev = None
|
||||
new_line =[]
|
||||
for y in x:
|
||||
this = y
|
||||
if prev != None:
|
||||
if this > 0 and prev > 0:
|
||||
new_line.append(0)
|
||||
else:
|
||||
new_line.append(y)
|
||||
else:
|
||||
new_line.append(y)
|
||||
prev = this
|
||||
now_block.append(new_line)
|
||||
return np.array(now_block)
|
||||
|
||||
print('Start convertion')
|
||||
for midi_file in os.listdir(MIDI_DIRECTORY):
|
||||
try:
|
||||
print('Reading file: {}'.format(midi_file))
|
||||
song = roll.Multitrack('{}/{}'.format(MIDI_DIRECTORY, midi_file))
|
||||
# no_drums_mt = roll.Multitrack(tempo=120.0, downbeat=[0, 96, 192, 288], beat_resolution=24)
|
||||
intruments_only = roll.Multitrack(tempo=120.0, beat_resolution=24)
|
||||
|
||||
for track in song.tracks:
|
||||
if track.is_drum == False:
|
||||
print(track.name, track.program)
|
||||
intruments_only.append_track(track=track, pianoroll=track.pianoroll)
|
||||
instrument_track = track.pianoroll
|
||||
|
||||
# plt.imshow(instrument_track[24*8:24*24].T)
|
||||
# plt.savefig('data/0_{}.png'.format(midi_file))
|
||||
|
||||
instrument_track = erase_note_lenth(instrument_track.T).T
|
||||
# plt.imshow(instrument_track[24*8:24*24].T)
|
||||
# plt.savefig('data/1_{}.png'.format(midi_file))
|
||||
|
||||
|
||||
# instruments = no_drums_mt.get_merged_pianoroll(mode='sum')
|
||||
|
||||
beats = floor( (instrument_track.shape[0] / MIDI_RESOLUTION) / BEAT_PER_BATCH) * BEAT_PER_BATCH
|
||||
notes_for_beats = beats * MIDI_RESOLUTION
|
||||
|
||||
print('beats: ', beats)
|
||||
samples_of_song = np.asarray(np.split(instrument_track[:notes_for_beats], beats))
|
||||
samples_of_song = samples_of_song.reshape(int(beats/BEAT_PER_BATCH),BEAT_PER_BATCH,96,128)
|
||||
|
||||
print('Converted samples: {}'.format(samples_of_song.shape))
|
||||
samples = np.concatenate([samples_of_song,samples], axis=0)
|
||||
np.savez_compressed(SAMPLES_DIRECTORY,samples)
|
||||
|
||||
except Exception as error:
|
||||
print('Convertion faild: {}'.format(error))
|
||||
pass
|
||||
|
||||
finally:
|
||||
print('Done!')
|
||||
BIN
project/__pycache__/settings.cpython-36.pyc
Normal file
BIN
project/__pycache__/settings.cpython-36.pyc
Normal file
Binary file not shown.
BIN
project/__pycache__/settings.cpython-37.pyc
Normal file
BIN
project/__pycache__/settings.cpython-37.pyc
Normal file
Binary file not shown.
BIN
project/data/autoencoder_model.h5
Normal file
BIN
project/data/autoencoder_model.h5
Normal file
Binary file not shown.
BIN
project/data/midi/Lenny Kravitz - Are You Gonna Go My Way.mid
Normal file
BIN
project/data/midi/Lenny Kravitz - Are You Gonna Go My Way.mid
Normal file
Binary file not shown.
BIN
project/data/midi/arctic_monkeys-505.mid
Normal file
BIN
project/data/midi/arctic_monkeys-505.mid
Normal file
Binary file not shown.
BIN
project/data/midi/red_hot_chili_peppers-cant_stop.mid
Normal file
BIN
project/data/midi/red_hot_chili_peppers-cant_stop.mid
Normal file
Binary file not shown.
BIN
project/data/output/generated_bar.npz
Normal file
BIN
project/data/output/generated_bar.npz
Normal file
Binary file not shown.
BIN
project/data/output/generated_midi.mid
Normal file
BIN
project/data/output/generated_midi.mid
Normal file
Binary file not shown.
BIN
project/data/output/pianoroll.png
Normal file
BIN
project/data/output/pianoroll.png
Normal file
Binary file not shown.
|
After Width: | Height: | Size: 8.4 KiB |
BIN
project/data/samples.npz
Normal file
BIN
project/data/samples.npz
Normal file
Binary file not shown.
26
project/generate.py
Normal file
26
project/generate.py
Normal file
@ -0,0 +1,26 @@
|
||||
import numpy as np
|
||||
import tensorflow as tf
|
||||
from keras.layers import Input, Dense, Conv2D
|
||||
from keras.models import Model
|
||||
import settings
|
||||
|
||||
#model
|
||||
input_shape = settings.midi_resolution*128
|
||||
input_img = tf.keras.layers.Input(shape=(input_shape,))
|
||||
encoded = tf.keras.layers.Dense(160, activation='relu')(input_img)
|
||||
decoded = tf.keras.layers.Dense(input_shape, activation='sigmoid')(encoded)
|
||||
autoencoder = tf.keras.models.Model(input_img, decoded)
|
||||
|
||||
autoencoder.compile(optimizer='adadelta',
|
||||
loss='categorical_crossentropy',
|
||||
metrics=['accuracy'])
|
||||
|
||||
# load weights into new model
|
||||
autoencoder.load_weights(settings.model_path)
|
||||
print("Loaded model from {}".format(settings.model_path))
|
||||
|
||||
# generate_seed = np.random.rand(12288).reshape(1,12288)
|
||||
generate_seed = np.load(settings.samples_path)['arr_0'][15].reshape(1,12288)
|
||||
|
||||
generated_sample = autoencoder.predict(generate_seed)
|
||||
np.savez_compressed(settings.generated_sample_path, generated_sample)
|
||||
249
project/midi_to_samples.py
Normal file
249
project/midi_to_samples.py
Normal file
@ -0,0 +1,249 @@
|
||||
import settings
|
||||
import pypianoroll as roll
|
||||
import matplotlib.pyplot as plt
|
||||
import numpy as np
|
||||
import os
|
||||
from tqdm import tqdm
|
||||
from math import floor
|
||||
import sys
|
||||
from sklearn.preprocessing import MinMaxScaler
|
||||
|
||||
midi_program = {
|
||||
0 : 'Perc',
|
||||
1 : 'Acoustic Grand Piano',
|
||||
2 : 'Bright Acoustic Piano',
|
||||
3 : 'Electric Grand Piano',
|
||||
4 : 'Honky-tonk Piano',
|
||||
5 : 'Electric Piano 1',
|
||||
6 : 'Electric Piano 2',
|
||||
7 : 'Harpsichord',
|
||||
8 : 'Clavi',
|
||||
9 : 'Celesta',
|
||||
10 : 'Glockenspiel',
|
||||
11 : 'Music Box',
|
||||
12 : 'Vibraphone',
|
||||
13 : 'Marimba',
|
||||
14 : 'Xylophone',
|
||||
15 : 'Tubular Bells',
|
||||
16 : 'Dulcimer',
|
||||
17 : 'Drawbar Organ',
|
||||
18 : 'Percussive Organ',
|
||||
19 : 'Rock Organ',
|
||||
20 : 'Church Organ',
|
||||
21 : 'Reed Organ',
|
||||
22 : 'Accordion',
|
||||
23 : 'Harmonica',
|
||||
24 : 'Tango Accordion',
|
||||
25 : 'Acoustic Guitar (nylon)',
|
||||
26 : 'Acoustic Guitar (steel)',
|
||||
27 : 'Electric Guitar (jazz)',
|
||||
28 : 'Electric Guitar (clean)',
|
||||
29 : 'Electric Guitar (muted)',
|
||||
30 : 'Overdriven Guitar',
|
||||
31 : 'Distortion Guitar',
|
||||
32 : 'Guitar harmonics',
|
||||
33 : 'Acoustic Bass',
|
||||
34 : 'Electric Bass (finger)',
|
||||
35 : 'Electric Bass (pick)',
|
||||
36 : 'Fretless Bass',
|
||||
37 : 'Slap Bass 1',
|
||||
38 : 'Slap Bass 2',
|
||||
39 : 'Synth Bass 1',
|
||||
40 : 'Synth Bass 2',
|
||||
41 : 'Violin',
|
||||
42 : 'Viola',
|
||||
43 : 'Cello',
|
||||
44 : 'Contrabass',
|
||||
45 : 'Tremolo Strings',
|
||||
46 : 'Pizzicato Strings',
|
||||
47 : 'Orchestral Harp',
|
||||
48 : 'Timpani',
|
||||
49 : 'String Ensemble 1',
|
||||
50 : 'String Ensemble 2',
|
||||
51 : 'SynthStrings 1',
|
||||
52 : 'SynthStrings 2',
|
||||
53 : 'Choir Aahs',
|
||||
54 : 'Voice Oohs',
|
||||
55 : 'Synth Voice',
|
||||
56 : 'Orchestra Hit',
|
||||
57 : 'Trumpet',
|
||||
58 : 'Trombone',
|
||||
59 : 'Tuba',
|
||||
60 : 'Muted Trumpet',
|
||||
61 : 'French Horn',
|
||||
62 : 'Brass Section',
|
||||
63 : 'SynthBrass 1',
|
||||
64 : 'SynthBrass 2',
|
||||
65 : 'Soprano Sax',
|
||||
66 : 'Alto Sax',
|
||||
67 : 'Tenor Sax',
|
||||
68 : 'Baritone Sax',
|
||||
69 : 'Oboe',
|
||||
70 : 'English Horn',
|
||||
71 : 'Bassoon',
|
||||
72 : 'Clarinet',
|
||||
73 : 'Piccolo',
|
||||
74 : 'Flute',
|
||||
75 : 'Recorder',
|
||||
76 : 'Pan Flute',
|
||||
77 : 'Blown Bottle',
|
||||
78 : 'Shakuhachi',
|
||||
79 : 'Whistle',
|
||||
80 : 'Ocarina',
|
||||
81 : 'Lead 1 (square)',
|
||||
82 : 'Lead 2 (sawtooth)',
|
||||
83 : 'Lead 3 (calliope)',
|
||||
84 : 'Lead 4 (chiff)',
|
||||
85 : 'Lead 5 (charang)',
|
||||
86 : 'Lead 6 (voice)',
|
||||
87 : 'Lead 7 (fifths)',
|
||||
88 : 'Lead 8 (bass + lead)',
|
||||
89 : 'Pad 1 (new age)',
|
||||
90 : 'Pad 2 (warm)',
|
||||
91 : 'Pad 3 (polysynth)',
|
||||
92 : 'Pad 4 (choir)',
|
||||
93 : 'Pad 5 (bowed)',
|
||||
94 : 'Pad 6 (metallic)',
|
||||
95 : 'Pad 7 (halo)',
|
||||
96 : 'Pad 8 (sweep)',
|
||||
97 : 'FX 1 (rain)',
|
||||
98 : 'FX 2 (soundtrack)',
|
||||
99 : 'FX 3 (crystal)',
|
||||
100 : 'FX 4 (atmosphere)',
|
||||
101 : 'FX 5 (brightness)',
|
||||
102 : 'FX 6 (goblins)',
|
||||
103 : 'FX 7 (echoes)',
|
||||
104 : 'FX 8 (sci-fi)',
|
||||
105 : 'Sitar',
|
||||
106 : 'Banjo',
|
||||
107 : 'Shamisen',
|
||||
108 : 'Koto',
|
||||
109 : 'Kalimba',
|
||||
110 : 'Bag pipe',
|
||||
111 : 'Fiddle',
|
||||
112 : 'Shanai',
|
||||
113 : 'Tinkle Bell',
|
||||
114 : 'Agogo',
|
||||
115 : 'Steel Drums',
|
||||
116 : 'Woodblock',
|
||||
117 : 'Taiko Drum',
|
||||
118 : 'Melodic Tom',
|
||||
119 : 'Synth Drum',
|
||||
120 : 'Reverse Cymbal',
|
||||
121 : 'Guitar Fret Noise',
|
||||
122 : 'Breath Noise',
|
||||
123 : 'Seashore',
|
||||
124 : 'Bird Tweet',
|
||||
125 : 'Telephone Ring',
|
||||
126 : 'Helicopter',
|
||||
127 : 'Applause',
|
||||
128 : 'Gunshot'
|
||||
}
|
||||
|
||||
# strasznie wolna funcja ;/
|
||||
def trim_notes(pianoroll):
|
||||
now_block = []
|
||||
for x in pianoroll:
|
||||
this = None
|
||||
prev = None
|
||||
new_line =[]
|
||||
for y in x:
|
||||
this = y
|
||||
if prev != None:
|
||||
if this > 0 and prev > 0:
|
||||
new_line.append(0)
|
||||
else:
|
||||
new_line.append(y)
|
||||
else:
|
||||
new_line.append(y)
|
||||
prev = this
|
||||
now_block.append(new_line)
|
||||
return np.array(now_block)
|
||||
|
||||
def metrum_check(midi_lenght, metrum=4, beat_resolution=24):
|
||||
return True if midi_lenght % (metrum * beat_resolution) == 0 else False
|
||||
|
||||
# >>> data = [[-1, 2], [-0.5, 6], [0, 10], [1, 18]]
|
||||
# >>> scaler = MinMaxScaler()
|
||||
# >>> print(scaler.fit(data))
|
||||
# MinMaxScaler(copy=True, feature_range=(0, 1))
|
||||
# >>> print(scaler.data_max_)
|
||||
# [ 1. 18.]
|
||||
# >>> print(scaler.transform(data))
|
||||
# [[0. 0. ]
|
||||
# [0.25 0.25]
|
||||
# [0.5 0.5 ]
|
||||
# [1. 1. ]]
|
||||
# >>> print(scaler.transform([[2, 2]]))
|
||||
# [[1.5 0. ]]
|
||||
|
||||
|
||||
def to_samples(midi_file_path, midi_res=settings.midi_resolution, ignore_note_lenght=settings.ignore_note_lenght):
|
||||
print('exporting samples from: {}'.format(midi_file_path))
|
||||
midi_file = roll.Multitrack(midi_file_path)
|
||||
samples = None
|
||||
all_samples = np.empty((0,settings.midi_resolution,128))
|
||||
for track in midi_file.tracks:
|
||||
# if not track.is_drum:
|
||||
if not metrum_check(track.pianoroll.shape[0]):
|
||||
print('Track skipped')
|
||||
continue
|
||||
else:
|
||||
instrument_track = track.pianoroll
|
||||
instrument_track = trim_notes(instrument_track.T).T if ignore_note_lenght else instrument_track
|
||||
scaler = MinMaxScaler()
|
||||
instrument_track = scaler.fit_transform(instrument_track)
|
||||
whole_beats = int(instrument_track.shape[0] / midi_res)
|
||||
samples = instrument_track.reshape(whole_beats, midi_res, 128)
|
||||
print('Exported {} samples of {}'.format(whole_beats, midi_program[track.program]))
|
||||
all_samples = np.concatenate([samples, all_samples], axis=0)
|
||||
return all_samples
|
||||
|
||||
def to_midi(samples, output_path=settings.generated_midi_path, program=0, tempo=120, beat_resolution=settings.beat_resolution):
|
||||
tracks = [roll.Track(samples, program=program)]
|
||||
return_midi = roll.Multitrack(tracks=tracks, tempo=tempo, downbeat=[0, 96, 192, 288], beat_resolution=beat_resolution)
|
||||
roll.write(return_midi, settings.generated_midi_path)
|
||||
|
||||
def to_png(samples, output_path=settings.generated_pianoroll_path, horizontal=True):
|
||||
img = samples.T if horizontal else samples
|
||||
plt.imshow(img, cmap='gray')
|
||||
plt.savefig(output_path)
|
||||
|
||||
def delete_empty_samples(sample_pack):
|
||||
temp_sample_pack = sample_pack
|
||||
index_manipulator = 1
|
||||
for index, sample in enumerate(sample_pack):
|
||||
if sample.sum() == 0:
|
||||
temp_sample_pack = np.delete(temp_sample_pack, index-index_manipulator, axis=0)
|
||||
index_manipulator = index_manipulator + 1
|
||||
print('Deleted {} empty samples'.format(index_manipulator-1))
|
||||
return temp_sample_pack
|
||||
|
||||
def main():
|
||||
if sys.argv[1]=='-e':
|
||||
print('Exporting started...')
|
||||
sample_pack = np.empty((0,settings.midi_resolution,128))
|
||||
for midi_file in os.listdir(settings.midi_dir):
|
||||
midi_file_path = '{}/{}'.format(settings.midi_dir, midi_file)
|
||||
midi_samples = to_samples(midi_file_path)
|
||||
if midi_samples is None:
|
||||
continue
|
||||
sample_pack = np.concatenate((midi_samples, sample_pack), axis=0)
|
||||
|
||||
sample_pack = delete_empty_samples(sample_pack)
|
||||
np.savez_compressed(settings.samples_dir, sample_pack)
|
||||
print('Exported {} samples'.format(sample_pack.shape[0]))
|
||||
elif sys.argv[1]=='-c':
|
||||
to_midi(settings.generated_sample_path)
|
||||
print('Samples to midi saved to {}'.format(settings.generated_sample_path))
|
||||
elif sys.argv[1]=='-p':
|
||||
sample_pack = np.load(settings.samples_path)['arr_0']
|
||||
for i, sample in tqdm(enumerate(sample_pack)):
|
||||
to_png(sample, output_path='data/preview/{}.png'.format(i))
|
||||
if i>50:
|
||||
sys.exit()
|
||||
else:
|
||||
print('type command afrer -e to export samples, -c to convert samples to midi, -p to preview samples in png')
|
||||
|
||||
if __name__ == '__main__':
|
||||
main()
|
||||
18
project/samples_to_midi.py
Normal file
18
project/samples_to_midi.py
Normal file
@ -0,0 +1,18 @@
|
||||
import pypianoroll as roll
|
||||
import matplotlib.pyplot as plt
|
||||
import numpy as np
|
||||
import os
|
||||
import settings
|
||||
|
||||
instruments = np.load(settings.generated_sample_path)['arr_0'][0]
|
||||
|
||||
instruments = instruments.reshape(96,128)
|
||||
# instruments = instruments>0.5
|
||||
instruments = instruments*255
|
||||
|
||||
i = roll.Track(instruments, program=0)
|
||||
generated_midi = roll.Multitrack(tracks=[i], tempo=120.0, downbeat=[0, 96, 192, 288], beat_resolution=24)
|
||||
roll.write(generated_midi, settings.generated_midi_path)
|
||||
|
||||
plt.imshow(instruments.T, cmap='gray')
|
||||
plt.savefig(settings.generated_pianoroll_path)
|
||||
17
project/settings.py
Normal file
17
project/settings.py
Normal file
@ -0,0 +1,17 @@
|
||||
# paths
|
||||
midi_dir = 'data/midi'
|
||||
samples_dir = 'data/samples'
|
||||
samples_path = 'data/samples.npz'
|
||||
model_path = 'data/autoencoder_model.h5'
|
||||
generated_sample_path = 'data/output/generated_bar.npz'
|
||||
generated_midi_path = 'data/output/generated_midi.mid'
|
||||
generated_pianoroll_path = 'data/output/pianoroll.png'
|
||||
|
||||
# export_settings
|
||||
midi_resolution = 96
|
||||
beat_resolution = 24
|
||||
beats_per_sample = 1
|
||||
ignore_note_lenght = False
|
||||
|
||||
#train_settings
|
||||
epochs = 1000
|
||||
31
project/train.py
Normal file
31
project/train.py
Normal file
@ -0,0 +1,31 @@
|
||||
import tensorflow as tf
|
||||
import settings
|
||||
from tensorflow.keras import layers
|
||||
from keras.layers import Input, Dense, Conv2D, Flatten
|
||||
from keras.models import Model, Sequential
|
||||
import numpy as np
|
||||
from sys import exit
|
||||
import pickle
|
||||
|
||||
print('Reading samples from: {}'.format(settings.samples_path))
|
||||
|
||||
train_X = np.load(settings.samples_path)['arr_0']
|
||||
|
||||
n_samples = train_X.shape[0]
|
||||
input_shape = settings.midi_resolution*128
|
||||
train_X = train_X.reshape(n_samples, input_shape)
|
||||
|
||||
# encoder model
|
||||
input_img = tf.keras.layers.Input(shape=(input_shape,))
|
||||
encoded = tf.keras.layers.Dense(160, activation='relu')(input_img)
|
||||
decoded = tf.keras.layers.Dense(input_shape, activation='sigmoid')(encoded)
|
||||
autoencoder = tf.keras.models.Model(input_img, decoded)
|
||||
|
||||
autoencoder.compile(optimizer='adam',
|
||||
loss='binary_crossentropy',
|
||||
metrics=['accuracy'])
|
||||
|
||||
autoencoder.fit(train_X, train_X, epochs=settings.epochs, batch_size=32)
|
||||
|
||||
autoencoder.save_weights(settings.model_path)
|
||||
print("Model save to {}".format(settings.model_path))
|
||||
@ -1,26 +0,0 @@
|
||||
GENERATED_BEAT_PATH = 'data/output/generated_bar.npz'
|
||||
OUTPUT_PATH = 'data/output/generated_midi.mid'
|
||||
OUTPUT_PATH_PIANOROLL = 'data/output/pianoroll.png'
|
||||
|
||||
import pypianoroll as roll
|
||||
import matplotlib.pyplot as plt
|
||||
import numpy as np
|
||||
import os
|
||||
|
||||
instruments = np.load(GENERATED_BEAT_PATH)['arr_0'][0]
|
||||
|
||||
instruments = instruments.reshape(96,128)
|
||||
instruments = instruments>instruments.min()*0.3
|
||||
instruments = instruments*255
|
||||
|
||||
# zeros_up = np.zeros((instruments.shape[0],24))
|
||||
# zeros_down = np.zeros((instruments.shape[0], 20))
|
||||
# instruments_full = np.concatenate([zeros_up,instruments], axis=1)
|
||||
# instruments_full = np.concatenate([instruments_full,zeros_down], axis=1)
|
||||
|
||||
i = roll.Track(instruments, program=0)
|
||||
return_midi = roll.Multitrack(tracks=[i], tempo=120.0, downbeat=[0, 96, 192, 288], beat_resolution=24)
|
||||
roll.write(return_midi, OUTPUT_PATH)
|
||||
|
||||
plt.imshow(instruments.T, cmap='gray')
|
||||
plt.savefig(OUTPUT_PATH_PIANOROLL)
|
||||
@ -1,4 +0,0 @@
|
||||
midi_path = 'data/midi'
|
||||
samples_path = 'data/samples'
|
||||
midi_resolution = 96
|
||||
beats_per_sample = 1
|
||||
38
train.py
38
train.py
@ -1,38 +0,0 @@
|
||||
SAMPLES_PATH = 'data/samples.npz'
|
||||
MODEL_PATH = 'data/autoencoder_model.h5'
|
||||
EPOCHS = 100
|
||||
|
||||
import tensorflow as tf
|
||||
from tensorflow.keras import layers
|
||||
from keras.layers import Input, Dense, Conv2D, Flatten
|
||||
from keras.models import Model
|
||||
import numpy as np
|
||||
from sys import exit
|
||||
import pickle
|
||||
|
||||
print('Reading samples from: {}'.format(SAMPLES_PATH))
|
||||
|
||||
train_samples = np.load(SAMPLES_PATH)['arr_0']
|
||||
train_samples = train_samples.reshape(train_samples.shape[0], 1*96*128)
|
||||
# input = Input(shape=(1,96,128))
|
||||
# encoded = Conv2D(filters = 32, kernel_size = 1, activation='relu')(input)
|
||||
# decoded = Conv2D(filters = 128, kernel_size = 1, activation='sigmoid')(encoded)
|
||||
# autoencoder = Model(input, decoded)
|
||||
#
|
||||
# autoencoder.compile(optimizer='adadelta',
|
||||
# loss='binary_crossentropy',
|
||||
# metrics=['accuracy'])
|
||||
|
||||
|
||||
encoded = Dense(128, input_shape=(1*96*128))
|
||||
decoded = Dense(96*128)(encoded)
|
||||
autoencoder = Model(input,decoded)
|
||||
|
||||
autoencoder.compile(optimizer='adadelta',
|
||||
loss='binary_crossentropy',
|
||||
metrics=['accuracy'])
|
||||
|
||||
autoencoder.fit(train_samples, train_samples, epochs=EPOCHS, batch_size=150)
|
||||
|
||||
autoencoder.save_weights(MODEL_PATH)
|
||||
print("Saved model to disk")
|
||||
Loading…
Reference in New Issue
Block a user