Difference between revisions of "MIR workshop 2008 notes"

From CCRMA Wiki
Jump to: navigation, search
(Timing and Segmentation)
(Tempo Extraction)
Line 17: Line 17:
 
** [http://www.iro.umontreal.ca/~pift6080/documents/papers/scheirer_jasa.pdf Tempo and beat analysis of acoustic musical signals]
 
** [http://www.iro.umontreal.ca/~pift6080/documents/papers/scheirer_jasa.pdf Tempo and beat analysis of acoustic musical signals]
 
** [http://staff.aist.go.jp/m.goto/PAPER/JNMR2001goto.pdf An Audio-based Real-time Beat Tracking System for Music With or Without Drum-sounds]*
 
** [http://staff.aist.go.jp/m.goto/PAPER/JNMR2001goto.pdf An Audio-based Real-time Beat Tracking System for Music With or Without Drum-sounds]*
*
 
  
 
= Feature Extraction =
 
= Feature Extraction =

Revision as of 10:40, 12 July 2010

This page is intended to supplement the lecture material found in the class - providing extra tutorials, support, references for further reading, or demonstration code snippets for those interested in a given topic. Please contribute to this growing list of resources. Do you have a great explanation of how a technique works? Found a great Java applet that illustrates a concept? Discovered a great survey of the field for a particular area? Please add it for the benefit of future students. Thanks!

I encourage you to ADD links and sections - but please do not REMOVE headings or items from the page.

Contents

Timing and Segmentation

Onset Detection

  • Papers:
  • Code:

Beat Extraction

  • Papers:
  • Code:

Tempo Extraction

Feature Extraction

Low Level Features

Zero Crossing, Temporal centroid, Log Attack time, Attack slope), Spectral features (Centroid, Flux, RMS, Rolloff, Flatness, Kurtosis, Brightness),Spectral bands, Log spectrogram

Chroma bins

MFCC

Auditory Toolbox (code and docs)

MPEG-7

Higher-level features

Key Estimation

Chord Estimation

Genre (genre, artist ID, similarity)

"Fingerprints"

Visualizing and Sonifying Feature data

Matt Hoffman's feature sonification work

Analysis / Decision Making

Classification

Heuristic Analysis

Distance measures (Euclidean, Manhattan, etc.)

k-NN

SVM / One-class SVM

Resources

Code

Clustering and probability density models

Density distance measures (centroid distance, EMD, KL-divergence, etc)

k-Means

Clustering

GMM

HMM

Nested classifier / Anchor-space / template-based systems

Model / Data Preparation Techniques

Data Preparation

PCA / LDA

Scaling data

Model organization

  • concept, design, data set construction and organization

Evaluation Methodology

Feature selection

Cross Validation

Information Retrieval metrics (precision, recall, F-Measure)

Real-world applications

Audio Segmentation

Automatic Audio Segmentation: Segment Boundary and Structure Detection in Popular Music

Audio Fingerprinting

The Last.fm fingerprinter uses this approach, code can be checked out from: svn://svn.audioscrobbler.net/recommendation/MusicID/lastfm_fplib

Drum Transcription

Audio Similarity

Music Recommendation / Playlisting

Getting Involved in the MIR Community

Research Databases / Collections of Ground truth data and copyright-cleared music


General MIR Datasets


Download links for the ISMIR 2004 genre classification contest training set:


Tags:

More:


From Georg Holzmann: 
LIST OF PUBLIC AVAILABLE MIR DATASETS
Downloadable Datasets:
- University of Iowa musical instruments samples:
   http://theremin.music.uiowa.edu/MIS.html
   Instrument samples recorded by the University of Iowa

- ISMIR2004 Audio Description Contest Dataset:
   http://ismir2004.ismir.net/ISMIR_Contest.html
   Datasets for
   - Genre Classification/Artist Identification
   - Melody Extraction
   - Tempo Induction
   - Rhythm Classification

- Graham's Melody Extraction Dataset:
   http://www.ee.columbia.edu/~graham/mirex_melody/
   http://labrosa.ee.columbia.edu/projects/melody/
   Audio files with correspondig pitch data

- MIREX06 Audio Tempo Extraction and Beat Tracking Datasets:
 
http://www.music-ir.org/mirex/2006/index.php/Audio_Tempo_Extraction#Practice_Data

- QBSH: A Corpus for Designing QBSH (Query by Singing/Humming) Systems
   http://neural.cs.nthu.edu.tw/jang2/dataSet/childSong4public/QBSH-corpus/

- Uni Dortmund Music Audio Benchmark Data Set:
   http://www-ai.cs.uni-dortmund.de/audio.html
   Songs from different genres and with tags (from garageband.com)

- Latin Music Database:
   http://www.ppgia.pucpr.br/~silla/lmd/
   3.160 music pieces in MP3 Format classified in 10 diferent musical genres
   (only features online)


Orderable Datasets:
- RWC Music Database:
   http://staff.aist.go.jp/m.goto/RWC-MDB/
   (many CDs)
   Datasets for
   - Pop Music & Royalty-Free Music
   - Classical Music
   - Jazz Music
   - Music Genre
   - Musical Instrument Sound

   Additional: AIST RWC Annotations
   http://staff.aist.go.jp/m.goto/RWC-MDB/AIST-Annotation/
   Additional annotations to the RWC database (beat, melody, ...)

- McGill University Master Samples:
   http://www.music.mcgill.ca/resources/mums/html/
   3 DVDs with instrument samples

- USPOP2002 Pop Music data set:
   http://labrosa.ee.columbia.edu/projects/musicsim/uspop2002.html
   (3 DVDs)
   MFCC features from 706 albums and 8764 tracks (400 artists)
   with style tags

- ENST-Drums:
   http://perso.telecom-paristech.fr/~gillet/ENST-drums/
   An extensive audio-visual database for drum signals processing


Free Online Music:

- magnatune.com creative commons music:
   http://magnatune.com/info/press/coverage/ccblog

- http://www.garageband.com/
   Public domain recordings

- http://epitonic.com/
   "high quality free and legal mp3 music"

- http://www.jamendo.com/
   Creative commons licensed music

- http://musicbrainz.org/
   Get music metadata

- http://www.freesound.org/
   Collaborative database of Creative Commons licensed sounds
   (not focused on songs)

Webservices:
- Networked Environment for Music Analysis:
   http://nema.lis.uiuc.edu/
   A webservices system for submitting code, running it against virtual collections
   (full use in 2010)

- MIREX DIY Framework:
   http://www.music-ir.org/mirexdiy/
   http://www.dlib.org/dlib/december06/downie/12downie.html
   (useable ?)

MIR Software and Toolboxes

Incomplete but growing list (courtesy of Joern Loviscach):
* MARSYAS
* jAudio
* Chuck
* The Sonic Visualizer/Annotator
* CLAM
* Music-to-Knowledge (M2K)
* MIRtoolbox
* MA toolbox
* Psysound
* Praat
* IPEM
* EchoNest
* libxtract
* MuBu
* Soundspotter
* timbreID
* openSMILE
* MPEG-7 XM
* MPEG-7 Audio Encoder
* MPEG-7 Audio Analyzer
* Sphinx 4 - Java-based open-source speech recognizer  http://cmusphinx.sourceforge.net/sphinx4/#capabilities

MIR Topic Areas

From Simon Dixon, Music-IR list, Dec 2008.

MIR Systems
- Content-based Querying
- Classification (genre/style/mood)
- Recommendation / playlist generation
- Fingerprinting / DRM
- Score following / Audio alignment
- Transcription / Annotation
- Tempo induction / Beat tracking
- Summarisation
- Streaming
- Text/web mining
- Optical music recognition
- Database systems / indexing / query languages

Human issues
- user interfaces, user models
- emotion, aesthetics
- perception, cognition
- social issues
- legal and ethical issues
- business issues
- methodological and philosophical issues

Data and metadata
- audio
- MIDI
- score
- text/web
- KR schemes, standards and protocols
- libraries and collections
- test sets and evaluation

Musical knowledge
- Melody and motives
- Harmony, chords and tonality
- Rhythm, beat, tempo and form
- Timbre, instrumentation and voice
- Genre, style and mood
- Performance
- Composition
- Ethnomusicology