MusicBrainz Summit/13

Summary

MusicBrainz Summit 13 occurred on the 21st and 22nd of September, 2013 at Wikimedia Germany in Berlin.

Quote of the summit: "Can you please toss me the broccoli?"

Sponsors

A big thank you to Google and Spotify for sponsoring the 2013 summit, and to Wikimedia Deutschland for hosting and providing good internet!

Summit attendees

Special guests:

• Lydia Pintscher from Wikimedia / Wikidata
• Anders Arpteg from Spotify

MusicBrainz community:

• CatCat
• fractalizator
• Freso
• ianmcorvidae
• ijabz
• JonnyJD
• kepstin
• LordSputnik
• Mineo
• navap
• nikki
• ocharles
• reosarevok
• ruaok
• warp

Agenda

Saturday
11 - 12	Introductions Setting the agenda State of the MB Spotify introduction (Anders Arpteg)
12 - 1pm	Geordi introduction External data matching
1 - 1:30	Break
1:30 - 1:45	Spotify ingestion process
1:45 - 2:15	NES Overview
2:30 - 2:45	Improving UX
2:45 - 3	Guiding new users (edit visibility, voting, subscriptions, etc.)
3 - 3:20	Break
3:20 - 3:30	Artist/label art
3:30 - 4	Scrobbling Acoustic data
4 - 4:20	Voting
4:20 - 4:40	Break
4:40 - 5:30	Genres

Sunday
10:30 - 10:40	Agenda revision
10:40 - 11	i18n
11 - 11:20	Promotion ideas
11:20 - 11:40	DiscIDs
11:40 - 12:10	Break
12:10 - 12:30	Home page redesign
12:30 - 12:50	Series
12:50 - 1:00	Elephant IDs
1 - 2	Lunch
2 - 2:05	Classical
2:05 - 2:15	WavePlot
2:15 - 2:30	MusicBrainz Audio Player
2:30 - 5	NES & UX breakout sessions Socializing

Discussion notes

Saturday

State of the MB

• MB is still hosted in California; hosting on autopilot, “when it dies, it dies”. Basic support from hosting folks to keep things running.
• Long term plans are to eventually move hosting away from California
• Online music monetization
  • Zoe Keating is an example of an artist making real money online and she publishes her revenue statements
  • Artists/labels are using YouTube as a monetization strategy (https://docs.google.com/spreadsheet/ccc?key=0AkasqHkVRM1OdEJFUnhyNFFkZjVSUWxhWGl1dE9lQXc)
  • Would be nice to leverage YouTube to build better label relationships

Geordi

• Development led by Ian
• Geordi is a third party dataset ingester
• Data mappings are created per third party dataset
• Data is then made available at https://geordi.musicbrainz.org for users to complete the ingestion

Spotify

• Spotify uses data from official label sources as well as third party sources
• They would like to explore a (better) relationship with MusicBrainz data
• They would like to have the “perfect music for every moment”

MusicBottle / User friendliness

• MusicBottle - Freso hasn’t had time to work on it lately, so on hold for now
• kepstin has been working on an alternative front end http://mbjs.kepstin.ca/. Source code repository at: https://github.com/kepstin/mbjs

Video

• Checkbox to be added to recordings marking them as video as a “stopgap” minimum measure
• Intent is to add a relationship from these “video” recordings to a third party video database

NES (New Edit System) Overview

• Development led by Ollie
• NES introduces a new edit system comprised of a front end that submits to a backend editable web service
• NES will allow bundling multiple edits together (but not unbundling, at least initially)
• Breakout session on Sunday

Improving UX

• The intent is not to dumb the site down, but to add improvements to the existing UX
• Proposal: “Fix bits that stick out”
  • Locate individual editing features that behave strangely, or don’t “do the right thing” by default, and perform minimal fixes.
• Proposal: “Multiple modes”
  • An analogy: with scanning software, you often have a “Simple” mode which lets you scan with some common settings, and “Advanced mode”, where you can specify descreening, colour curves, etc.
• Proposal: “Better defaults and default-hidden advanced features”
  • Instead of full modes, just reorder things so simple/important stuff first and possibly hide features that many users might not need (with a preference to always show them)

Guiding new users

• Proposal: Create a music dashboard that lists new users that need help
• Proposal: Create multiple subscription lists
  • Having multiple lists narrows the scope of each list and makes it easier to digest

Artist/label art

• Now that CAA has been a success, intent is to expand and also cover artist and label images
• Proposal: Use a whitelist of CC-licensed websites and start displaying the art
  • Whitelist: WikiMedia Commons, Flickr, and more
  • For Flickr and other API-enabled websites, ModBot will add a comment with the license and where it comes from

Scrobbling

• Proposal: Set up a scrobbling server under the MetaBrainz umbrella
• The initial API would be last.fm compatible
• Over the long term the intent is to improve the API and build support for it directly into third party apps (eg. players) to increase user base

Acoustic data

• Essentia does cool shit! The proposal is to collect data output from it and make it available to whoever wants to use it for recommendation, etc.
• UPF can run it on the research dataset put out by the internet archive as some seed data, and it could potentially be integrated with a tool like Picard/Jaikoz, similar to how AcoustID collection works.
• WavePlot already does similar things and might constitute a starting point for a database of such data

Voting

• Intent is to increase the voting being performed in the database
• Previously (~6 years ago) we had an iframe at the top of the page, this resulted in a lot of ‘no’ votes and was seen as an annoyance
• Proposal: Remove ‘Yes’ votes
• Proposal: Display relevant edit information to improve voting workflow
• Proposal: Add ability to save custom searches

Genres

• 2012 discussion summary:
  1. Multi level list of genres
     • Soundunwound solution: “main”, “quite”, “hints of”
     • 2 levels: “main genre”, “sub genre(s)”
  2. Flat list of genres
     • Fixed list to choose from
     • Free form with autocompletion
  3. Genre graph
• Ben’s “genre entity” proposal
  • Genres are entered by users as freeform, auto-completed tags
  • After reaching a certain threshold, the tag is converted into a “genre entity”
  • Displayed genres are comprised of both the tags and genre entities
  • Converting a genre tag into a genre entity allows it to have relationships, and a genre entity would have fields for id3/itunes/(spotify?) genre mappings
  • Tags can be manually promoted to entities with a vote
  • Genre entities can be merged. The title of the merge source becomes a search hint for the merge target
  • This would allow things like “Pop-Rock” and “poprock” to be merged and typing either of these as tags would result in the same genre entity being applied to the entity.
• No conclusion here. It might be good to do some genres meetings, because this is a feature we should have and it would be good to have a better plan/timeframe for implementing it.

Sunday

Internationalization

• Discussion point: whether to enable translation support on the main site
  • Proposal: Enable all languages that are on beta on main
  • Proposal: Enable just one language to start
• Discussion held on whether to enable other languages even though all the documentation is all in English
• Consensus reached that waiting for the all the documentation to be translated into other language(s) is not going to work
• Consensus reached that we enable just German (most Germans speak English) and then go from there
• Can we look into translating the wiki? Wikimedia DE has had good experience with Extension:Translate. The documentation team would also like this.

Promotion ideas

• First impression for new users is lacking (home page needs work)
• Developer awareness can be raised by promoting MB-aware apps such as those created as a part of Music Hack Day
• Build a promotion-oriented community of users to coordinate promotional releases (reddit posts, social media, etc.), collaborate via a new mailing list
• A possible Summer of Code project is to create a visually pleasing statistics page similar to VGMdb

DiscIDs

• Different ideas discussed:
  • DiscIDs prevent a few useful edge cases like pre-gap tracks and correct tracklists for certain VideoCDs.
  • But they’re also used a lot by rippers/taggers, so we cannot just get rid of them.
• Decision reached that we look at our DiscID usage statistics and then revisit this topic
• There is a follow-up in User:JonnyJD/DiscID with details.

Home page redesign ideas

• short desc/tagline
  • "an open music encyclopedia"?
• divide people into categories in as few words as possible.
  • general users / people who want to tag
  • artists / labels
  • devs
• show coverart
• show “recent activity” in an activity steam
• very basic stats
• highlight search bar?

Series

• Consensus that everyone wants series support
  • New entity should be created with
    • Title
    • Sub entities (see proposals below)
    • Relationships to labels, artists
    • Dates derived from contents
    • Ordering of sub entities is optional
• Consensus split on what can be put into a series
  • Proposal: Just release groups
  • Proposal: Either release groups or releases, but not both at the same time
  • Proposal: One of the above AND recordings (podcasts being captured via stand alone recordings)
• Motion carried to prototype and test the workability of a series entity that contains just release groups

Elephant IDs

• aka adding track IDs into music files
• Motion carried to add the MusicBrainz Track ID into a tag called "musicbrainz_releasetrackid"

Classical

• Consensus reached that classical discussion requires a separate summit
• Further discussion tabled

WavePlot

• Development on WavePlot led by Ben
• WavePlot is a system for making images of audio files and extracting data from then.
• Future work will be improving integration with existing tools and adding features.

MusicBrainz Audio Player

• Consensus reached that the optimal audio player is the MusicBrainz website itself
• Proposal is to further develop keptsin’s MBJS and then integrate either a third party streaming service, a web based player, or a desktop player
• Consensus reached that an optimal player also needs to be accessible/usable entirely offline.