SlideShare a Scribd company logo
1 of 30
Improving perceptual tempo estimation
   with crowd-sourced annotations
        Mark Levy, 26 October 2011
Tempo Estimation
Terminology:
 tempo = beats per minute = bpm
Tempo Estimation
Use crowd-sourcing:
 quantify influence of metrical ambiguity

  on tempo perception
 improve evaluation


 improve algorithms
Perceived Tempo
Metrical ambiguity:
 listeners don’t agree about bpm


 typically in two camps


 perceived values differ by factor of 2 or 3




McKinney and Moelants:
 24-40 subjects


 released experimental data
Perceived Tempo
            Metrical ambiguity:
listeners




                                  listeners

                     bpm                      bpm

             McKinney and Moelants, 2004
Machine-Estimated Tempo
Also affected by metrical ambiguity:
 makes estimation difficult


 natural to see multiple bpm values


 estimated values often out by factor of 2 or 3

  (“octave error”)
Crowd Sourcing
Web-based questionnaire:
 capture label choices


 capture bpm from mean tapping interval


 capture comparative judgements
Crowd Sourcing
Crowd Sourcing
 Music:
  over 4000 songs


  30-second clips


• rock, country, pop, soul, funk and rnb, jazz,
   latin, reggae, disco, rap, punk, electronic,
   trance, industrial, house, folk, ...
• recent releases back to 60s
Response
First week (reported/released):
 4k tracks annotated by 2k listeners


 20k labels and bpm estimates




To date:
 6k tracks annotated by 27k listeners


 200k labels and bpm estimates
Analysis: ambiguity
When people tap to a song at different bpm
 do they really disagree about whether it’s

  slow or fast?

Investigation:
 inspect labels from people who tap differently


 quantify disagreement for ambiguous songs
Analysis: ambiguity
Subset of slow/fast songs:
 labelled by at least five listeners


 majority label “slow” or “fast”
Analysis: ambiguity
bpm vs speed label




all estimates for slow/fast songs
Analysis: ambiguity
bpm vs speed label



            people can tap slowly to fast songs




all estimates for slow/fast songs
Analysis: ambiguity
Labels for fast songs from slow-tappers
Analysis: ambiguity
Quantify disagreement over labels:
 model conflict, extremity of tempo


 conflict coefficient


               min(Ls , L f ) Ls       Lf
          C
               max(Ls , L f )      L

  Ls, Lf, L: number of slow, fast, all labels for a song
Analysis: ambiguity
Distribution of conflict coefficient C




            C > 0 means slow and fast


all songs with at least five labels
Analysis: ambiguity
Subset of metrically ambiguous songs:
 at least 30% of listeners tap at half/twice the

  majority estimate

Compared to the rest:
 no significant difference in C
Evaluation metrics
MIREX:
 capture metrical ambiguity


 replicate human disagreement




Ambiguity considered unhelpful:
 automatic playlisting


 DJ tools, production tools


 jogging
Evaluation metrics
Application-oriented :
 compare with majority* human estimate
    (*median in most popular bin)
   categorise machine estimates
          same as humans
          twice as fast
          twice as slow
          three times as fast
          and so on
          unrelated to humans
Analysis: evaluation
Sources:
 BPM List (DJ kit, human-moderated)

    Donny Brusca, 7th edition, 2011
   EchoNest/MSD (closed-source algorithm)
    maybe Jehan et al,?
   VAMP (open-source algorithm)
    Davies and Landone, 2007-
Analysis: machine vs human
    80%

    70%

    60%

    50%
                                               BPM List
    40%
                                               VAMP
    30%                                        EchoNest

    20%

    10%

     0%
          x2   same   /2   unrelated   other
Analysis: controlled test
Controlled comparison:
 exploit experience from website A/B testing


 use this to improve algorithm iteratively




Result is independent of any quality metric
Analysis: controlled test
When visitor arrives at the page:
 choose a source S at random


 choose a bpm value at random


 choose two songs given that value by S


 display them together




Then ask which sounds faster!
Analysis: controlled test
Null Hypothesis:
 there will be presentation effects


 listeners will attend to subtle differences


but
 these effects are independent of the source

  of bpm estimates
 if the quality of the sources is the same
Analysis: controlled test
     100%
     90%
     80%
     70%
     60%
     50%                                 different
     40%                                 same

     30%
     20%
     10%
      0%
            BPM List   VAMP   EchoNest
Analysis: improving estimates
Adjust bpm based on class:
 imagine an accurate slow/fast classifier

       Hockmann and Fujinaga, 2010
   adjust as follows:
      bpm:= bpm/2 if slow and bpm > 100
      bpm:= bpm*2 if fast and bpm < 100
      otherwise don’t adjust
   simulation: accept majority human label
Analysis: adjusted vs human
    80%

    70%

    60%

    50%
                                               BPM List
    40%
                                               VAMP
    30%                                        EchoNest

    20%

    10%

     0%
          x2   same   /2   unrelated   other
Conclusions
Crowd sourcing:
 gather thousands of data points in a few

  days, half a million over time
 humans agree over slow/fast labels, even

  when they tap at different bpm
Improving machine estimates:
 use controlled testing


 exploit a slow/fast classifier
Thanks!
mark@last.fm      @gamboviol

http://mir-in-action.blogspot.com
http://playground.last.fm/demo/speedo
http://users.last.fm/~mark/speedo.tgz

We are looking for interns/research fellows!

More Related Content

Viewers also liked

Presentation Edge Amsterdam Crowdsourcing
Presentation Edge Amsterdam CrowdsourcingPresentation Edge Amsterdam Crowdsourcing
Presentation Edge Amsterdam CrowdsourcingDerek Ariëns
 
Crowdsourcing for Marketing Workshop - New York
Crowdsourcing for Marketing Workshop - New YorkCrowdsourcing for Marketing Workshop - New York
Crowdsourcing for Marketing Workshop - New YorkRoss Dawson
 
Crowdsourcing Creative Ideas
Crowdsourcing Creative IdeasCrowdsourcing Creative Ideas
Crowdsourcing Creative Ideasquirkemarketing
 
From Crowdsourcing to Crowd Making: The Path From Ideas to Solutions
From Crowdsourcing to Crowd Making: The Path From Ideas to SolutionsFrom Crowdsourcing to Crowd Making: The Path From Ideas to Solutions
From Crowdsourcing to Crowd Making: The Path From Ideas to SolutionsSeattle Interactive Conference
 
Crowdsourcing @ Migros
Crowdsourcing @ MigrosCrowdsourcing @ Migros
Crowdsourcing @ MigrosMigros
 
Put the Crowd to Work
Put the Crowd to WorkPut the Crowd to Work
Put the Crowd to WorkSteve Buttry
 
Building crowdsourcing applications
Building crowdsourcing applicationsBuilding crowdsourcing applications
Building crowdsourcing applicationsSimon Willison
 
Crowdsourcing-future-advertising
Crowdsourcing-future-advertisingCrowdsourcing-future-advertising
Crowdsourcing-future-advertisingDerek Ariëns
 
Managing the crowd- HBR article
Managing the crowd-  HBR articleManaging the crowd-  HBR article
Managing the crowd- HBR articleAbhishek kyal
 
Marketing lessons from Narendra Modi
Marketing lessons from Narendra ModiMarketing lessons from Narendra Modi
Marketing lessons from Narendra ModiSrinivas Chari
 
Crowd Sourcing And Humanistic Psychology
Crowd Sourcing And Humanistic PsychologyCrowd Sourcing And Humanistic Psychology
Crowd Sourcing And Humanistic PsychologyJW Alphenaar
 
Future of Crowdsourcing: Creation to Curation, Search to Synthesis, Content t...
Future of Crowdsourcing: Creation to Curation, Search to Synthesis, Content t...Future of Crowdsourcing: Creation to Curation, Search to Synthesis, Content t...
Future of Crowdsourcing: Creation to Curation, Search to Synthesis, Content t...Gaurav Mishra
 
Die Studierende von heute sind die Designer der Zukunft
Die Studierende von heute sind die Designer der ZukunftDie Studierende von heute sind die Designer der Zukunft
Die Studierende von heute sind die Designer der ZukunftZwetana Penova
 
Shut Up and Take My Money: LEGO Does Crowdsourcing
Shut Up and Take My Money: LEGO Does CrowdsourcingShut Up and Take My Money: LEGO Does Crowdsourcing
Shut Up and Take My Money: LEGO Does CrowdsourcingTim Courtney
 
101: Crowdsourcing vs Wisdom of the Crowd
101: Crowdsourcing vs Wisdom of the Crowd101: Crowdsourcing vs Wisdom of the Crowd
101: Crowdsourcing vs Wisdom of the Crowdpeter bihr
 
Crowdsourcing challenges and opportunities 2012
Crowdsourcing challenges and opportunities 2012Crowdsourcing challenges and opportunities 2012
Crowdsourcing challenges and opportunities 2012xin wang
 

Viewers also liked (18)

Presentation Edge Amsterdam Crowdsourcing
Presentation Edge Amsterdam CrowdsourcingPresentation Edge Amsterdam Crowdsourcing
Presentation Edge Amsterdam Crowdsourcing
 
Crowdsourcing for Marketing Workshop - New York
Crowdsourcing for Marketing Workshop - New YorkCrowdsourcing for Marketing Workshop - New York
Crowdsourcing for Marketing Workshop - New York
 
Crowdsourcing Creative Ideas
Crowdsourcing Creative IdeasCrowdsourcing Creative Ideas
Crowdsourcing Creative Ideas
 
Crowd Sourcing Platforms - An UnPresentation
Crowd Sourcing Platforms - An UnPresentationCrowd Sourcing Platforms - An UnPresentation
Crowd Sourcing Platforms - An UnPresentation
 
From Crowdsourcing to Crowd Making: The Path From Ideas to Solutions
From Crowdsourcing to Crowd Making: The Path From Ideas to SolutionsFrom Crowdsourcing to Crowd Making: The Path From Ideas to Solutions
From Crowdsourcing to Crowd Making: The Path From Ideas to Solutions
 
Crowd Sourcing
Crowd SourcingCrowd Sourcing
Crowd Sourcing
 
Crowdsourcing @ Migros
Crowdsourcing @ MigrosCrowdsourcing @ Migros
Crowdsourcing @ Migros
 
Put the Crowd to Work
Put the Crowd to WorkPut the Crowd to Work
Put the Crowd to Work
 
Building crowdsourcing applications
Building crowdsourcing applicationsBuilding crowdsourcing applications
Building crowdsourcing applications
 
Crowdsourcing-future-advertising
Crowdsourcing-future-advertisingCrowdsourcing-future-advertising
Crowdsourcing-future-advertising
 
Managing the crowd- HBR article
Managing the crowd-  HBR articleManaging the crowd-  HBR article
Managing the crowd- HBR article
 
Marketing lessons from Narendra Modi
Marketing lessons from Narendra ModiMarketing lessons from Narendra Modi
Marketing lessons from Narendra Modi
 
Crowd Sourcing And Humanistic Psychology
Crowd Sourcing And Humanistic PsychologyCrowd Sourcing And Humanistic Psychology
Crowd Sourcing And Humanistic Psychology
 
Future of Crowdsourcing: Creation to Curation, Search to Synthesis, Content t...
Future of Crowdsourcing: Creation to Curation, Search to Synthesis, Content t...Future of Crowdsourcing: Creation to Curation, Search to Synthesis, Content t...
Future of Crowdsourcing: Creation to Curation, Search to Synthesis, Content t...
 
Die Studierende von heute sind die Designer der Zukunft
Die Studierende von heute sind die Designer der ZukunftDie Studierende von heute sind die Designer der Zukunft
Die Studierende von heute sind die Designer der Zukunft
 
Shut Up and Take My Money: LEGO Does Crowdsourcing
Shut Up and Take My Money: LEGO Does CrowdsourcingShut Up and Take My Money: LEGO Does Crowdsourcing
Shut Up and Take My Money: LEGO Does Crowdsourcing
 
101: Crowdsourcing vs Wisdom of the Crowd
101: Crowdsourcing vs Wisdom of the Crowd101: Crowdsourcing vs Wisdom of the Crowd
101: Crowdsourcing vs Wisdom of the Crowd
 
Crowdsourcing challenges and opportunities 2012
Crowdsourcing challenges and opportunities 2012Crowdsourcing challenges and opportunities 2012
Crowdsourcing challenges and opportunities 2012
 

Similar to Crowd sourcing for tempo estimation

IRJET- Implementing Musical Instrument Recognition using CNN and SVM
IRJET- Implementing Musical Instrument Recognition using CNN and SVMIRJET- Implementing Musical Instrument Recognition using CNN and SVM
IRJET- Implementing Musical Instrument Recognition using CNN and SVMIRJET Journal
 
Tonic Identification System for Hindustani and Carnatic Music
Tonic Identification System for Hindustani and Carnatic MusicTonic Identification System for Hindustani and Carnatic Music
Tonic Identification System for Hindustani and Carnatic MusicSankalp Gulati
 
music recommendation system ,Based on Million Song Dataset
music recommendation system ,Based on Million Song Datasetmusic recommendation system ,Based on Million Song Dataset
music recommendation system ,Based on Million Song DatasetSandipKumarPratihari
 
MOODetector: A System for Mood-based Classification and Retrieval of Audio Mu...
MOODetector: A System for Mood-based Classification and Retrieval of Audio Mu...MOODetector: A System for Mood-based Classification and Retrieval of Audio Mu...
MOODetector: A System for Mood-based Classification and Retrieval of Audio Mu...Rui Pedro Paiva
 
web based music genre classification.pptx
web based music genre classification.pptxweb based music genre classification.pptx
web based music genre classification.pptxUmaMahesh786960
 

Similar to Crowd sourcing for tempo estimation (8)

IRJET- Implementing Musical Instrument Recognition using CNN and SVM
IRJET- Implementing Musical Instrument Recognition using CNN and SVMIRJET- Implementing Musical Instrument Recognition using CNN and SVM
IRJET- Implementing Musical Instrument Recognition using CNN and SVM
 
Tonic Identification System for Hindustani and Carnatic Music
Tonic Identification System for Hindustani and Carnatic MusicTonic Identification System for Hindustani and Carnatic Music
Tonic Identification System for Hindustani and Carnatic Music
 
music recommendation system ,Based on Million Song Dataset
music recommendation system ,Based on Million Song Datasetmusic recommendation system ,Based on Million Song Dataset
music recommendation system ,Based on Million Song Dataset
 
MOODetector: A System for Mood-based Classification and Retrieval of Audio Mu...
MOODetector: A System for Mood-based Classification and Retrieval of Audio Mu...MOODetector: A System for Mood-based Classification and Retrieval of Audio Mu...
MOODetector: A System for Mood-based Classification and Retrieval of Audio Mu...
 
Emofy
Emofy Emofy
Emofy
 
Song Proffer
Song ProfferSong Proffer
Song Proffer
 
About
AboutAbout
About
 
web based music genre classification.pptx
web based music genre classification.pptxweb based music genre classification.pptx
web based music genre classification.pptx
 

Recently uploaded

From Event to Action: Accelerate Your Decision Making with Real-Time Automation
From Event to Action: Accelerate Your Decision Making with Real-Time AutomationFrom Event to Action: Accelerate Your Decision Making with Real-Time Automation
From Event to Action: Accelerate Your Decision Making with Real-Time AutomationSafe Software
 
How to convert PDF to text with Nanonets
How to convert PDF to text with NanonetsHow to convert PDF to text with Nanonets
How to convert PDF to text with Nanonetsnaman860154
 
Handwritten Text Recognition for manuscripts and early printed texts
Handwritten Text Recognition for manuscripts and early printed textsHandwritten Text Recognition for manuscripts and early printed texts
Handwritten Text Recognition for manuscripts and early printed textsMaria Levchenko
 
04-2024-HHUG-Sales-and-Marketing-Alignment.pptx
04-2024-HHUG-Sales-and-Marketing-Alignment.pptx04-2024-HHUG-Sales-and-Marketing-Alignment.pptx
04-2024-HHUG-Sales-and-Marketing-Alignment.pptxHampshireHUG
 
Powerful Google developer tools for immediate impact! (2023-24 C)
Powerful Google developer tools for immediate impact! (2023-24 C)Powerful Google developer tools for immediate impact! (2023-24 C)
Powerful Google developer tools for immediate impact! (2023-24 C)wesley chun
 
Tata AIG General Insurance Company - Insurer Innovation Award 2024
Tata AIG General Insurance Company - Insurer Innovation Award 2024Tata AIG General Insurance Company - Insurer Innovation Award 2024
Tata AIG General Insurance Company - Insurer Innovation Award 2024The Digital Insurer
 
Presentation on how to chat with PDF using ChatGPT code interpreter
Presentation on how to chat with PDF using ChatGPT code interpreterPresentation on how to chat with PDF using ChatGPT code interpreter
Presentation on how to chat with PDF using ChatGPT code interpreternaman860154
 
Understanding Discord NSFW Servers A Guide for Responsible Users.pdf
Understanding Discord NSFW Servers A Guide for Responsible Users.pdfUnderstanding Discord NSFW Servers A Guide for Responsible Users.pdf
Understanding Discord NSFW Servers A Guide for Responsible Users.pdfUK Journal
 
Advantages of Hiring UIUX Design Service Providers for Your Business
Advantages of Hiring UIUX Design Service Providers for Your BusinessAdvantages of Hiring UIUX Design Service Providers for Your Business
Advantages of Hiring UIUX Design Service Providers for Your BusinessPixlogix Infotech
 
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...Miguel Araújo
 
2024: Domino Containers - The Next Step. News from the Domino Container commu...
2024: Domino Containers - The Next Step. News from the Domino Container commu...2024: Domino Containers - The Next Step. News from the Domino Container commu...
2024: Domino Containers - The Next Step. News from the Domino Container commu...Martijn de Jong
 
TrustArc Webinar - Stay Ahead of US State Data Privacy Law Developments
TrustArc Webinar - Stay Ahead of US State Data Privacy Law DevelopmentsTrustArc Webinar - Stay Ahead of US State Data Privacy Law Developments
TrustArc Webinar - Stay Ahead of US State Data Privacy Law DevelopmentsTrustArc
 
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024The Digital Insurer
 
Exploring the Future Potential of AI-Enabled Smartphone Processors
Exploring the Future Potential of AI-Enabled Smartphone ProcessorsExploring the Future Potential of AI-Enabled Smartphone Processors
Exploring the Future Potential of AI-Enabled Smartphone Processorsdebabhi2
 
A Call to Action for Generative AI in 2024
A Call to Action for Generative AI in 2024A Call to Action for Generative AI in 2024
A Call to Action for Generative AI in 2024Results
 
The 7 Things I Know About Cyber Security After 25 Years | April 2024
The 7 Things I Know About Cyber Security After 25 Years | April 2024The 7 Things I Know About Cyber Security After 25 Years | April 2024
The 7 Things I Know About Cyber Security After 25 Years | April 2024Rafal Los
 
What Are The Drone Anti-jamming Systems Technology?
What Are The Drone Anti-jamming Systems Technology?What Are The Drone Anti-jamming Systems Technology?
What Are The Drone Anti-jamming Systems Technology?Antenna Manufacturer Coco
 
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...Drew Madelung
 
[2024]Digital Global Overview Report 2024 Meltwater.pdf
[2024]Digital Global Overview Report 2024 Meltwater.pdf[2024]Digital Global Overview Report 2024 Meltwater.pdf
[2024]Digital Global Overview Report 2024 Meltwater.pdfhans926745
 
08448380779 Call Girls In Greater Kailash - I Women Seeking Men
08448380779 Call Girls In Greater Kailash - I Women Seeking Men08448380779 Call Girls In Greater Kailash - I Women Seeking Men
08448380779 Call Girls In Greater Kailash - I Women Seeking MenDelhi Call girls
 

Recently uploaded (20)

From Event to Action: Accelerate Your Decision Making with Real-Time Automation
From Event to Action: Accelerate Your Decision Making with Real-Time AutomationFrom Event to Action: Accelerate Your Decision Making with Real-Time Automation
From Event to Action: Accelerate Your Decision Making with Real-Time Automation
 
How to convert PDF to text with Nanonets
How to convert PDF to text with NanonetsHow to convert PDF to text with Nanonets
How to convert PDF to text with Nanonets
 
Handwritten Text Recognition for manuscripts and early printed texts
Handwritten Text Recognition for manuscripts and early printed textsHandwritten Text Recognition for manuscripts and early printed texts
Handwritten Text Recognition for manuscripts and early printed texts
 
04-2024-HHUG-Sales-and-Marketing-Alignment.pptx
04-2024-HHUG-Sales-and-Marketing-Alignment.pptx04-2024-HHUG-Sales-and-Marketing-Alignment.pptx
04-2024-HHUG-Sales-and-Marketing-Alignment.pptx
 
Powerful Google developer tools for immediate impact! (2023-24 C)
Powerful Google developer tools for immediate impact! (2023-24 C)Powerful Google developer tools for immediate impact! (2023-24 C)
Powerful Google developer tools for immediate impact! (2023-24 C)
 
Tata AIG General Insurance Company - Insurer Innovation Award 2024
Tata AIG General Insurance Company - Insurer Innovation Award 2024Tata AIG General Insurance Company - Insurer Innovation Award 2024
Tata AIG General Insurance Company - Insurer Innovation Award 2024
 
Presentation on how to chat with PDF using ChatGPT code interpreter
Presentation on how to chat with PDF using ChatGPT code interpreterPresentation on how to chat with PDF using ChatGPT code interpreter
Presentation on how to chat with PDF using ChatGPT code interpreter
 
Understanding Discord NSFW Servers A Guide for Responsible Users.pdf
Understanding Discord NSFW Servers A Guide for Responsible Users.pdfUnderstanding Discord NSFW Servers A Guide for Responsible Users.pdf
Understanding Discord NSFW Servers A Guide for Responsible Users.pdf
 
Advantages of Hiring UIUX Design Service Providers for Your Business
Advantages of Hiring UIUX Design Service Providers for Your BusinessAdvantages of Hiring UIUX Design Service Providers for Your Business
Advantages of Hiring UIUX Design Service Providers for Your Business
 
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
 
2024: Domino Containers - The Next Step. News from the Domino Container commu...
2024: Domino Containers - The Next Step. News from the Domino Container commu...2024: Domino Containers - The Next Step. News from the Domino Container commu...
2024: Domino Containers - The Next Step. News from the Domino Container commu...
 
TrustArc Webinar - Stay Ahead of US State Data Privacy Law Developments
TrustArc Webinar - Stay Ahead of US State Data Privacy Law DevelopmentsTrustArc Webinar - Stay Ahead of US State Data Privacy Law Developments
TrustArc Webinar - Stay Ahead of US State Data Privacy Law Developments
 
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
 
Exploring the Future Potential of AI-Enabled Smartphone Processors
Exploring the Future Potential of AI-Enabled Smartphone ProcessorsExploring the Future Potential of AI-Enabled Smartphone Processors
Exploring the Future Potential of AI-Enabled Smartphone Processors
 
A Call to Action for Generative AI in 2024
A Call to Action for Generative AI in 2024A Call to Action for Generative AI in 2024
A Call to Action for Generative AI in 2024
 
The 7 Things I Know About Cyber Security After 25 Years | April 2024
The 7 Things I Know About Cyber Security After 25 Years | April 2024The 7 Things I Know About Cyber Security After 25 Years | April 2024
The 7 Things I Know About Cyber Security After 25 Years | April 2024
 
What Are The Drone Anti-jamming Systems Technology?
What Are The Drone Anti-jamming Systems Technology?What Are The Drone Anti-jamming Systems Technology?
What Are The Drone Anti-jamming Systems Technology?
 
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
 
[2024]Digital Global Overview Report 2024 Meltwater.pdf
[2024]Digital Global Overview Report 2024 Meltwater.pdf[2024]Digital Global Overview Report 2024 Meltwater.pdf
[2024]Digital Global Overview Report 2024 Meltwater.pdf
 
08448380779 Call Girls In Greater Kailash - I Women Seeking Men
08448380779 Call Girls In Greater Kailash - I Women Seeking Men08448380779 Call Girls In Greater Kailash - I Women Seeking Men
08448380779 Call Girls In Greater Kailash - I Women Seeking Men
 

Crowd sourcing for tempo estimation

  • 1. Improving perceptual tempo estimation with crowd-sourced annotations Mark Levy, 26 October 2011
  • 2. Tempo Estimation Terminology:  tempo = beats per minute = bpm
  • 3. Tempo Estimation Use crowd-sourcing:  quantify influence of metrical ambiguity on tempo perception  improve evaluation  improve algorithms
  • 4. Perceived Tempo Metrical ambiguity:  listeners don’t agree about bpm  typically in two camps  perceived values differ by factor of 2 or 3 McKinney and Moelants:  24-40 subjects  released experimental data
  • 5. Perceived Tempo Metrical ambiguity: listeners listeners bpm bpm McKinney and Moelants, 2004
  • 6. Machine-Estimated Tempo Also affected by metrical ambiguity:  makes estimation difficult  natural to see multiple bpm values  estimated values often out by factor of 2 or 3 (“octave error”)
  • 7. Crowd Sourcing Web-based questionnaire:  capture label choices  capture bpm from mean tapping interval  capture comparative judgements
  • 9. Crowd Sourcing Music:  over 4000 songs  30-second clips • rock, country, pop, soul, funk and rnb, jazz, latin, reggae, disco, rap, punk, electronic, trance, industrial, house, folk, ... • recent releases back to 60s
  • 10. Response First week (reported/released):  4k tracks annotated by 2k listeners  20k labels and bpm estimates To date:  6k tracks annotated by 27k listeners  200k labels and bpm estimates
  • 11. Analysis: ambiguity When people tap to a song at different bpm  do they really disagree about whether it’s slow or fast? Investigation:  inspect labels from people who tap differently  quantify disagreement for ambiguous songs
  • 12. Analysis: ambiguity Subset of slow/fast songs:  labelled by at least five listeners  majority label “slow” or “fast”
  • 13. Analysis: ambiguity bpm vs speed label all estimates for slow/fast songs
  • 14. Analysis: ambiguity bpm vs speed label people can tap slowly to fast songs all estimates for slow/fast songs
  • 15. Analysis: ambiguity Labels for fast songs from slow-tappers
  • 16. Analysis: ambiguity Quantify disagreement over labels:  model conflict, extremity of tempo  conflict coefficient min(Ls , L f ) Ls Lf C max(Ls , L f ) L Ls, Lf, L: number of slow, fast, all labels for a song
  • 17. Analysis: ambiguity Distribution of conflict coefficient C C > 0 means slow and fast all songs with at least five labels
  • 18. Analysis: ambiguity Subset of metrically ambiguous songs:  at least 30% of listeners tap at half/twice the majority estimate Compared to the rest:  no significant difference in C
  • 19. Evaluation metrics MIREX:  capture metrical ambiguity  replicate human disagreement Ambiguity considered unhelpful:  automatic playlisting  DJ tools, production tools  jogging
  • 20. Evaluation metrics Application-oriented :  compare with majority* human estimate (*median in most popular bin)  categorise machine estimates  same as humans  twice as fast  twice as slow  three times as fast  and so on  unrelated to humans
  • 21. Analysis: evaluation Sources:  BPM List (DJ kit, human-moderated) Donny Brusca, 7th edition, 2011  EchoNest/MSD (closed-source algorithm) maybe Jehan et al,?  VAMP (open-source algorithm) Davies and Landone, 2007-
  • 22. Analysis: machine vs human 80% 70% 60% 50% BPM List 40% VAMP 30% EchoNest 20% 10% 0% x2 same /2 unrelated other
  • 23. Analysis: controlled test Controlled comparison:  exploit experience from website A/B testing  use this to improve algorithm iteratively Result is independent of any quality metric
  • 24. Analysis: controlled test When visitor arrives at the page:  choose a source S at random  choose a bpm value at random  choose two songs given that value by S  display them together Then ask which sounds faster!
  • 25. Analysis: controlled test Null Hypothesis:  there will be presentation effects  listeners will attend to subtle differences but  these effects are independent of the source of bpm estimates  if the quality of the sources is the same
  • 26. Analysis: controlled test 100% 90% 80% 70% 60% 50% different 40% same 30% 20% 10% 0% BPM List VAMP EchoNest
  • 27. Analysis: improving estimates Adjust bpm based on class:  imagine an accurate slow/fast classifier Hockmann and Fujinaga, 2010  adjust as follows: bpm:= bpm/2 if slow and bpm > 100 bpm:= bpm*2 if fast and bpm < 100 otherwise don’t adjust  simulation: accept majority human label
  • 28. Analysis: adjusted vs human 80% 70% 60% 50% BPM List 40% VAMP 30% EchoNest 20% 10% 0% x2 same /2 unrelated other
  • 29. Conclusions Crowd sourcing:  gather thousands of data points in a few days, half a million over time  humans agree over slow/fast labels, even when they tap at different bpm Improving machine estimates:  use controlled testing  exploit a slow/fast classifier
  • 30. Thanks! mark@last.fm @gamboviol http://mir-in-action.blogspot.com http://playground.last.fm/demo/speedo http://users.last.fm/~mark/speedo.tgz We are looking for interns/research fellows!