Slide 1
Slide 1 text
ਪ͠ͱ࢝ΊΔMIR
ͯʔͱ͘ (@tetoku_sakana)
2020-02-13 #stapy54
Slide 2
Slide 2 text
͜Μͳײ͡ͰਐΈ·͢
• ࣗݾհ
• MIRͱ
• MIRͷ࣮ྫհ
• MIRͷ࢝Ίํ
Slide 3
Slide 3 text
୭
• ͯʔͱ͘ (@tetoku_sakana)
• WebΤϯδχΞ
• nao_y ͞Μͷ͓༠͍ͰࢀՃͤ͞
͍ͯͨͩ͘͜ͱʹͳΓ·ͨ͠
Slide 4
Slide 4 text
No content
Slide 5
Slide 5 text
ਪ͠ࣄ ᶃ
• ΦαΧφϝʔλʔ
• @osakanameter
• ΦαΧφͷMVͷ࠶ੜճ
ϑΥϩϫʔͷՄ
ࢹԽ௨
• ެࣜϗʔϜϖʔδͷߋ৽
ใχϡʔεͷ৴
Slide 6
Slide 6 text
ਪ͠ࣄ ᶄ
• ΦαΧφΞʔΧΠϒ
• ΦαΧφʹ·ͭΘΔ
ΠϯλϏϡʔهࣄͳ
ͲΛ·ͱΊͨαʔϏ
ε
Slide 7
Slide 7 text
ͦΜͳ͜ΜͳͰ
MIRʹೖ͠·ͨ͠ʂ
Slide 8
Slide 8 text
MIR
Slide 9
Slide 9 text
No content
Slide 10
Slide 10 text
No content
Slide 11
Slide 11 text
MIR
• Music Information Retrieval
• ԻָใݕࡧԻָใॲཧͱ༁͞ΕΔ
• ݕࡧ͋Μ·Γؔͳ͍
• ػցֶश×Իָ
• (ओʹඇੜܥͷ) Իָؔ࿈ͷ૯শ
Slide 12
Slide 12 text
MIRͷλεΫ (Ұྫ)
• ԻָԻָใ (ௐςϯϙɺίʔυ) ͷݕ
ग़ɾਪఆ
• Իָͷࣗಈྨ (δϟϯϧงғؾͳͲ)
• ࣖίϐͷࣗಈԽ
• ԻָͰԻָΛݕࡧ
Slide 13
Slide 13 text
MIRͷख๏
• Content-based
• ԻָՎࢺͳͲɺָۂσʔλΛѻ͏
• Context-based
• ΞʔςΟετͷհจͳͲɺָۂҎ֎ͷपล
σʔλΛѻ͏
Slide 14
Slide 14 text
pythonͱMIR
• librosa
• madmom
• essentia
Slide 15
Slide 15 text
import librosa
>>> filepath = librosa.util.example_audio_file()
>>> y, sr = librosa.load(filepath, offset=30, duration=5)
>>> librosa.feature.mfcc(y=y, sr=sr) # MFCCͷऔಘ
array([[ -5.229e+02, -4.944e+02, ..., -5.229e+02, -5.229e+02],
[ 7.105e-15, 3.787e+01, ..., -7.105e-15, -7.105e-15],
...,
[ 1.066e-14, -7.500e+00, ..., 1.421e-14, 1.421e-14],
[ 3.109e-14, -5.058e+00, ..., 2.931e-14, 2.931e-14]])
Slide 16
Slide 16 text
ΦαΧφͷۂ͍͠…
Slide 17
Slide 17 text
ΦαΧφͷதͰ
Ұ൪ָ͍͠ۂʁ
Slide 18
Slide 18 text
ϚεϩοΫࢦ
Slide 19
Slide 19 text
No content
Slide 20
Slide 20 text
• Elias Pampalk et al., Proceedings of the ACM
Multimedia 2002
• ָۂྨࣅΛࣗݾ৫ԽϚοϓ(SOM)Λͬ
ͯՄࢹԽ
Content-based Organization and
Visualization of Music Archives
Slide 21
Slide 21 text
Իڹ৺ཧֶΛߟྀͨ͠ɺௌײ্ͷloudnessͷม
ԽΛಛྔͱͯ͠நग़͢Δ
“Rhythm Patterns”ͱͯ͠ఏҊ͞ΕͯΔಛྔநग़ख๏
Slide 22
Slide 22 text
No content
Slide 23
Slide 23 text
ϚεϩοΫࢦ͕
ࢉग़Ͱ͖ͦ͏ʂ
Slide 24
Slide 24 text
ॲཧ֓ཁ
1. STFTΛ͔͚ͯ(ରईͷ) εϖΫτϩάϥϜ
Λऔಘ
2. (1) ΛϒϩοΫʹׂ͠(rolling window)ɺͦ
ΕͧΕ࣌ؒ࣠ํʹSTFTΛ͔͚ͯέϓετϩ
άϥϜΛऔಘ
3. (2) ʹରͯ͠60ύʔηϯλΠϧΛٻΊΔ
Slide 25
Slide 25 text
No content
Slide 26
Slide 26 text
No content
Slide 27
Slide 27 text
def minmax(pattern):
return (pattern - pattern.min()) / (pattern.max() -
pattern.min())
# ࡶ
def mathrock_index(pattern):
pattern = pattern.sum(axis=0)
pattern = minmax(pattern) * 100
pattern = np.diff(pattern)
return np.percentile(pattern, q=90)
def calc_lfp(filename):
cent = CentSpectrum(win_length=2048, hop_length=512)
D = librosa.amplitude_to_db(cent.proc(filename))
D_normalized = cent.normalize(D)
lfp = LogarithmicFluctuationPattern(hop_length=256)
return lfp.proc(D_normalized)
Slide 28
Slide 28 text
No content
Slide 29
Slide 29 text
MIRͷ࢝Ίํ
Slide 30
Slide 30 text
MIRͷ࢝Ίํ
• MIREXISMIRͷจ
• ipynb
• musicinformationretrieval.com
• ΟʔϯՊେͷnbviewer
Slide 31
Slide 31 text
MIRͷ࢝Ίํ
• SpotifyͷAPIΛ͏
• Audio Features for a Track
• Audio Analysis for a Track
Slide 32
Slide 32 text
{
"danceability": 0.735,
"energy": 0.578,
"key": 5,
"loudness": -11.84,
"mode": 0,
"speechiness": 0.0461,
"acousticness": 0.514,
"instrumentalness": 0.0902,
"liveness": 0.159,
"valence": 0.624,
"tempo": 98.002,
"type": "audio_features",
"id": "06AKEBrKUckW0KREUWRnvT",
"uri": "spotify:track:06AKEBrKUckW0KREUWRnvT",
"track_href": “https://api.spotify.com/v1/tracks/…",
"analysis_url": “https://api.spotify.com/v1/audio-analysis/…”,
"duration_ms": 255349,
"time_signature": 4
}
Slide 33
Slide 33 text
No content
Slide 34
Slide 34 text
·ͱΊ
• ใগͳ͍͚ͲMIRͷෑډ͍
• ϥΠϒϥϦlibrosa͕͓͢͢Ί
• SpotifyͷAPI͓͢͢Ί
• ΦαΧφྑ͍
Slide 35
Slide 35 text
ਪ͠ۦಈ։ൃ
Ұॹʹ࢝ΊͯΈ·͠ΐ͏
͝੩ௌ͋Γ͕ͱ͏͍͟͝·ͨ͠ʙ