Teaching:TUW - UE InfoVis WS 2005/06 - Gruppe G4 - Aufgabe 3 - Design

Topic

MP3 Archive Visualization

Specification of the Application Area and the given Dataset

Application area Analysis

The application area in this case is the creation of a clearly arranged visualization for a big music archive consisting of several thousand files. This can be achieved by using the values of the ID3-tags as well as the attributes of the music files itself.

Since the existing data mainly consist of discrete values, our mission to produce an appropriate visualization will basically be achieved with the creation of quantitative data by using a reasonable combination of given values.

Most Audio formats nowadays use the container format ID3 to store additional Information about the file, apart from the given file properties.

ID3 generally supports loads of different input values and to visualize them all would go beyond the scope of our task. Besides in most cases only the most common values like Album, Interpret, Year… are specified correctly. Therefore we will concentrate our work for our prototype on these entries.

Dataset Analysis

Attribute	Data type	Description
Title	Discrete	Name of the Song
Interpret	Discrete	Name of the Artist/Band who plays the song
Album	Discrete	Name of the Album/Compilation the song belongs to
Year	Ordinal	Year the song was recorded in
Genre	Nominal	Genre the song belongs to
Size	Quantitative	Size of the file
Duration	Quantitative	Overall play length of the song
Bit rate	Ordinal	Bit rate the file was encoded with
Sample rate	Ordinal	Sample rate the file was encoded with
File Location	Discrete	Full system path to the file

A complete data record consists of all attributes listed above.

All data in this set is one-dimensional.

Analysis of the Target User Group

Who should use this kind of visualization technique?

This visualization technique is mainly meant for the 'end-users', that is someone who collects lots of MP3s. With 'lots of MP3s' we mean quite a few GBs, just more than 30 GBs. Our visualization should help the user to get an overview of his collection and his listening-habits. By using this visualization technique, the user will get informations to one chosen artist and whose discography. For example: in his database the user has got the band radiohead, who produced albums over 15 years and in this period they changed their musicstyle from alternative rock to experimental electronic. The visualization will show him from which producing period he has got more MP3s and which period he likes more, by counting the number of listenings of each song. The result could be that he has got more MP3s from their early years, but likes the experimental electronic tracks more. This visualization technique could also be interesting for the band and the music industry, if they want to produce a new album. for example: a band (like radiohead) who changed their musicstyle over the years and wants to know which style is preferd more. But therefor they have to compare these dates from many users.

What are the characteristics of the target group?

People of this group are music enthusiasts. They have thousands of mp3s on their harddisk and love it to collect them. Most of them have lost track of their collection, on the strength of the abundance of their collection. These people mainly receive their mp3s from the internet instead of buying CDs, because they like to see their whole musiccollection at a glance.

Are there any known / often used Methods / Visualisation Techniques?

No, until now only the textbased listing-technique is used.

Intended Purpose of our Visualization

What should be achieved with this visualization?

A better information representation of the mp3s should be achieved by combining several ID3-tags. The representation of the data and the contained information should be effective, precise and self-explanatory, so that the user can get a good and expressive overview of his collection and the contained information.

Which tasks should be solved?

The data readout of contentspecific characteristics should be possible. We want to visualize the distribution of the genres, so that the user is able to group them how he likes. e.g.: there is a group called 'classics' and the user wants to put the group 'oldies' to them, so he can make a subgroup.

Questions that should be solved with this visualization technique

It should be possible to get the frequency distribution of the mp3s relating to their characteristics. But the precondition therefor is that the ID3-tags are available

Proposal of Design

Types of Visualization applied

As there are no comparable features implemented in existing mp3 archival storage systems we decided to implement a visualization kit exploring the quantitative distribution of an mp3-archive according to the different genres. As a default the main window shows several genres the authors considered to be the right ones for music main-genres in the root directory including their percentage-rate corresponding to the whole file-archive.

By a single click on one of the main genres the user gets additional information in the upper right window on the one hand and of course a file-listing in the lower left window on the other hand. Double clicking a genre in the main window brings the user straight to a sub-level containing the corresponding sub-genres again shown in the main window. Through the path shown on top of the main window the system allows to easily jump back to wherever necessary out of every level.

Further on the user is able to read out file-specific information like Title, Interpret, Album, Year, Duration & Bit-Rate through a click on the particular file listed below, but is also allowed to save a play-list from the files selected via the file-explorer.

Following feature turns the visualization kit into a rather complex and outstanding tool. By default the tool sorts the input data through a given hierarchy according to the main genres and alternatively to the sub-genres following the ID3-Tags. Additionally the User has the ability to change this system and to create his own view according to his point of interest. That means that he is able to change the main genres and the sub-genres as well. As there are always different views especially on the terms of sub-genres the user is even allowed to change the structure of the given categories. That means that one is permitted to extract an ambiguous sub-genre to directly attach it to another genre. This will be able via the right middle window showing the sub-genres and their actual belongings as well. Probably the system will give the possibility to cancel unmeant genres at all.

Visual Mapping

• Due to the enormous number of music-genres ID3-Tags provide we will probably choose a tree structure in addition to the text-based genre-navigator on the right in lower levels. This structure should be similar to a hyperbolic tree.

Dimension "Genre-Multitude" --> visual attribute "Tree Branches"

• To afford best visualization we are right now not sure whether we should display the percentage-rate in bars as shown in the fake-screenshot below, or should rather use a Bubble-Chart, where the size of the bubbles is determined by the values of occurrence per genre.

Dimension "File-Occurrence" --> visual attribute "Area"

Specification of used Techniques / applied Principles

• Hyperbolic Trees:

(s. slide 18, 19, 20 of Info_Vis4.pdf handed out in the course “188.305 VO Informationsvisualisierung”)

Due to the enormous number of music-genres ID-3 Tags provide we probably choose a tree structure similar to a hyperbolic tree to visualize the raw data extracted from the corresponding mp3-archiv.

• Bubble Chart:

(s. similar to bubble-maps, slide 52 of Info_Vis4.pdf handed out in the course 188.305 VO Informationsvisualisierung, http://peltiertech.com/Excel/ChartsHowTo/HowToBubble.html)

Bubble chart, where the size of the bubbles is determined by the values of occurrence.

• Details on demand:

(s. slide 55 of Info_Vis4.pdf handed out in the course 188.305 VO Informationsvisualisierung)

Used several times. (File-Explorer, Genre-Information, ...)

• Linking & Brushing:

(s. slide 100 of Info_Vis0.pdf handed out in the course 188.305 VO Informationsvisualisierung)

Detail Windows containing percentage-distribution according to the specific genre

• Focus & Context: Tiled Multi-Level Browser

(slide 69 of Info_Vis0.pdf handed out in the course 188.305 VO Informationsvisualisierung)

Overview Window, Details on Demand Window

Possibilities of User-Interaction

• Assemble information about occurence-percentage per genre
• Assemble information according to ID3-Tags
• Group genres
• Build paths
• Build new tree
• Modify leaves
• Save play-list

Mockup / Fake Screenshot

References

[Wikipedia, 2005a] ID3, Wikipedia, Last updated: 21 November, 2005, Retrieved at: November 22, 2005, http://www.csam.montclair.edu/~mcdougal/SCP/D_types.htm

[Wikipedia, 2005b] MP3, Wikipedia, Last updated: 21 November, 2005, Retrieved at: November 22, 2005, http://en.wikipedia.org/wiki/Mp3

[Id3.org, 2004] ID3v2 frames, Id3.org, Last updated: 28. Februar, 2004, Retrieved at: November 22, 2005, http://www.id3.org/frames.html

[Montclaire, 2000] Data Types, Department of Science and Mathematics at Montclair State University, Last updated: 3. August, 2000, Retrieved at: November 22, 2005, http://www.csam.montclair.edu/~mcdougal/SCP/D_types.htm