Music Information Retrieval, or how to search for (and maybe find) music and do away with incipits - PowerPoint PPT Presentation

Loading...

PPT – Music Information Retrieval, or how to search for (and maybe find) music and do away with incipits PowerPoint presentation | free to view - id: ccaea-ZDc1Z



Loading


The Adobe Flash plugin is needed to view this content

Get the plugin now

View by Category
About This Presentation
Title:

Music Information Retrieval, or how to search for (and maybe find) music and do away with incipits

Description:

A tool to catalog and extract audio CD contents for online distribution. Automatic identification of CDs. Compute CDDB of the CD. Metadata retrieval and correction ... – PowerPoint PPT presentation

Number of Views:238
Avg rating:3.0/5.0
Slides: 20
Provided by: michaelf82
Category:

less

Write a Comment
User Comments (0)
Transcript and Presenter's Notes

Title: Music Information Retrieval, or how to search for (and maybe find) music and do away with incipits


1
Music Information Retrieval, or how to search for
(and maybe find) music and do away with incipits
  • Michael Fingerhut
  • Multimedia Library and Engineering Bureau
  • IRCAM Centre Pompidou

2
Why Music Information Retrieval
Implication
Technical Advance
Increased availability of musical contents in
digital form (locally)
Storage
Faster methods for processing contents and
producing "meaning"
Computing power
Increased availability of musical contents in
digital form (remotely)
Networks
Need for
... actions, methods and procedures for
recovering stored data to provide information on
music.
3
A (much simplified) MIR map
user
rights owner
cognitive, social
taste, mood
DRM
interaction
annotation
musicology
playlists
analysis
similarity
metadata
synchronization, summarization
theory
genre
indexation
pattern extraction, form recognition
information
semantic
motives
structure
features
feature extraction,identification
musicalacoustical
concept ? sign ? signal ? sign ?
concept
characteristics

symbolic
fingerprint
pitch, voice extraction
data
metadata
automatic
textual
sound
text
score
stored data
about
lyrics
symbolic
digital
OCR
paper
physical
performance (live, recorded)
publications
music
librarian
work
abstract
performer
author
composer
4
A typology of MIR
  • Preprocessing
  • OCR, digitization, compression
  • Encoding, notation
  • Feature extraction
  • Segmentation
  • Instrument recognition
  • Voice recognition
  • Indexation
  • Identification
  • Clustering
  • Classification
  • Extraction
  • Melody, key, harmony, rhythm
  • Structural analysis
  • Polyphony
  • Repetition
  • Similarity
  • Summarization
  • Organization
  • Search
  • Objective criteria
  • Metadata indices (name, title, period, genre,
    instrumentation)
  • Full-text (with or without semantic tags)
  • Query by example (audio excerpt, melody, contour,
    rhythm, tonality, harmony)
  • Similarity
  • Acoustical characteristics
  • Subjective criteria
  • Mood
  • Taste
  • Retrieve, deliver, use
  • Browsing
  • Playlists
  • Using and reusing (annotate, combine, transform)
  • Rights management (recognition, watermarking)
  • Usability
  • Evaluation
  • User studies

5
Common methods in MIR
  • Modeling
  • Start from a theory
  • Look for patterns
  • Statistical
  • Look for patterns
  • Build a theory
  • Evaluation
  • Relevance (recall, precision)
  • Performance (speed, friendliness)
  • Standardized test collections

query
Recall 30 (300 out of 1000) Precision 75
(300 out of 400)
6
MIR as a multidisciplinary domain
  • 000 General
  • 000 Generalities computer science
  • 001 Knowledge
  • 001.4 Research
  • 003 Systems
  • 004 Data processing
  • 004.5 Storage
  • 004.6 Interfacing and communications
  • 005 Computer programming
  • 005.4 Systems programming and programs
  • 005.7 Data in computer systems 005.8 Data
    security
  • 006 Special computer methods 
  • 006.3 Artificial intelligence
  • 006.4 Computer pattern recognition
  • 006.5 Computer sound synthesis
  • 020 Library and information sciences
  • 025 Library operations
  • 025.3 Bibliographic analysis and control
  • 025.4 Subject analysis and control
  • 300 Social sciences
  • 300 Sociology and anthropology
  • 302.2 Communication
  • 306 Culture and Institutions
  • 310 General Statistics
  • 340 Law
  • 341.7 Law of international cooperation
  • 380 Commerce
  • 384 Communications Telecommunications 
  • 500 Natural Science and Mathematics
  • 510 Mathematics
  • 516 Geometry
  • 600 General Technology
  • 620 Engineering allied operations
  • 620.2 Sound and related vibrarions
  • 621.3 Electric, electronic, magnetic,
    communications, computer engineering
  • 681 Precision instruments other devices
  • 700 The Arts
  • 780 Music

7
Take One, or SummarizingDave Brubecks Take Five
(5m25s)
1. Find which parts sound like other parts
(timbre similarity)
2. Extract segments
6 5 4 3 2 1
0m0s
5m25s
3. Produce summaries
all 14735 A small bit of each segment
each 01482 One segment of each type
longest 00325 Longest segment
most freq. 00325 Most frequent segment
0m0s
5m25s
similarity matrix The warmer the color (red
warmest), the more similar
Courtesy of Geoffroy Peeters, Ircam
8
A tool to catalog and extract audio CD contents
for online distribution
  • Automatic identification of CDs
  • Compute CDDB of the CD
  • Metadata retrieval and correction
  • Query Internet CDDB for metadata
  • Allow correction
  • Extraction and compression
  • Transfer to a Web server

9
(No Transcript)
10
(No Transcript)
11
(No Transcript)
12
(No Transcript)
13
(No Transcript)
14
(No Transcript)
15
(No Transcript)
16
(No Transcript)
17
(No Transcript)
18
(No Transcript)
19
Whats the use
  • Increased collections
  • Quantity
  • Variety
  • Help in organizing
  • Cataloguing and indexing
  • Help in finding
  • Search and retrieval
  • Contribute knowledge
  • Metadata, indexation
  • Relevance
  • Users
  • Getting involved
  • ISMIR annual conferenceswww.ismir.net
  • Be informed
  • Publish, speak
  • Review
  • Organize
About PowerShow.com