POPULARITY OF MUSIC PLAYER PARAMETERS AND FEATURESSkip other details (including permanent urls, DOI, citation information)
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License. Please contact email@example.com to use this work in a way not covered by the license. :
For more information, read Michigan Publishing's access and usage policy.
Page 1 ï~~POPULARITY OF MUSIC PLAYER PARAMETERS AND FEATURES Jukka Holm Nokia Research Center P. O. Box 1000 FIN-33721 Tampere Finland ABSTRACT One attractive alternative for a music player user interface is interactive visualization that facilitates browsing, organizing, and selecting what kind of music to listen to. To design this type of visualizations, we have arranged several online questionnaires on how people map graphical objects to musical properties such as genre and tempo. In order to find the most important parameters for our future studies, we arranged a questionnaire on the popularity of 71 selected music player parameters and features. We also studied how the participants selected music from their computers and mobile phones. In addition to music visualization, the results can be useful for the designers of more traditional music player UIs. 1. INTRODUCTION There are various ways to use a music player application. If the user knows what he wants to listen to, he can quickly locate the desired song or album and start the playback. The user may also browse through his collection in a various ways, the simplest being the alphabetically ordered list. However, as it is often easier to locate the desired music from a smaller set of songs, the music collection may also be reordered or filtered using some parameters such as genre, release year, tempo, user's own ratings for songs, and listening history. These parameters may also be used for searching a specific song or album. One popular way to select music is to generate playlists by selecting content song by song from the music collection. As manual playlist generation takes some time, the user can also use so-called smart playlists that are created automatically and contain only songs that match certain criteria (e.g. "add a song to the playlist, if its release year is less than 1990, it belongs to the genre of rock, and the tempo is faster than 120 bpm"). Yet another popular way to select music is to use shuffle (random) play mode in which the songs are played in a random order. Especially in the case of large music collections, an attractive alternative is interactive visualization that facilitates browsing, organizing, and selecting what kind of music to listen to. The purpose of music visualization is to replace or complement the traditional UI with a dynamic, adaptive, and interactive image that is based on one or more attributes of the songs. For example, the user could click on a sharp black object to start the playback of fast heavy metal from his music collection, or a blurred blue object would start the playback of old blues music. In a way, this type of visualizations can be seen as an easy way to generate smart playlists. To design the visualizations, we have previously arranged several online questionnaires on how people map graphical objects to musical properties such as genre, tempo , and release year . This paper describes the results of our latest questionnaire, concentrating on the popularity of 71 selected music player parameters and features. The purpose was to find out which of parameters and features are important to the participants and in which ways the participants typically use their music player software. The results will be taken into account when designing our forthcoming music visualization questionnaires, but they should also be useful for the designers of more traditional music players. 2. ONLINE QUESTIONNAIRE The call for participation was sent to c. 200 employees of a large international company. Participation was voluntary and the participants did not receive any compensation for participating in the study. 47 persons answered the questionnaire, and 85% of them were male and 15% female. 72 % (32 persons) were Finnish, and the rest British, Chinese, French, German, Indian, or Mexican. All participants except one were 25-45 years old, and majority of them had engineering background. 38.3% of participants listened to digital music with a computer, iPod, mobile phone, or other mobile music player several times a day. 29.8% did it once a day, 25.5% a couple of times of week, and the rest less frequently. The number of albums in participants' digital music collections varied quite a bit: 21.3% owned 0-50 albums, 14.9% 50-100, 2.1% 100-200, 27.7% 200-400, 14.9% 400-600, 8.5% 600-1000, 4.2% 1000-2000, and 6.4% had more than 2000. In the beginning of the questionnaire, the participants read a short introduction about the different ways of using a music player (browsing, filtering, generating playlists, etc.). After this, they were shown a list of 71 music player related parameters and features. The participants were asked to mark which ones they would like to use in a fictional music player for any purpose including those mentioned in the introduction. In addition, there were also some questions related to how the participants usually select songs from their own music players.
Page 2 ï~~3. RESULTS 3.1. Individual songs, albums, artists, and genres The first part of the questionnaire consisted of 12 song related parameters (Table 1). The participants were able to select one from the following options: A = I would like to use this parameter often B = I could use this parameter sometimes C = I would never use this parameter D = No opinion As seen from the table, the majority of participants were interested in using every listed parameter either often or sometimes. This implies that these parameters have deserved their place in commercial music player applications. In fact, while Windows Media Player  does not currently support tempo nor skip count, the other parameters can be found from both Apple iTunes  and Media Player. As expected, the most popular parameter was "artist name" (100% of participants voted for A or B) and the least popular "composer name" (51.1% A or B). Parameter A B C D Artist name 83.0 17.0 0.0 0.0 Song name 74.5 23.4 2.1 0.0 Composer name 8.5 42.6 42.5 6.4 Tempo 8.5 57.5 34.0 0.0 Genre 34.0 51.1 14.9 0.0 Length 21.3 34.0 40.4 4.3 Release year 10.6 44.7 36.2 8.5 My own rating 23.4 51.1 25.5 0.0 Play count 12.8 72.3 14.9 0.0 Skip count 10.6 53.2 29.8 6.4 Date last played 10.6 55.3 27.7 6.4 Date added 8.5 57.5 31.9 2.1 Table 1. Percentages of song parameter votes. Next, the participants were shown a list of 10 album related parameters (Table 2). Again, the majority of participants were interested in using them either often or sometimes. The only exception was album length, which most would never use. Unsurprisingly, the most popular parameter was "album name". Parameter A B C D Album name 74.5 19.1 6.4 0.0 Album cover art 40.4 40.4 17.0 2.2 Genre 19.1 66.0 12.8 2.1 Length 0.0 36.2 57.4 6.4 Release year 12.8 51.1 27.6 8.5 My own rating 14.9 57.4 21.3 6.4 Automatic rating 6.4 61.7 27.7 4.2 (based on the average of album's song ratings) Play count 8.5 55.3 23.4 12.8 Date last played 6.4 48.9 31.9 12.8 Date added 8.5 51.1 34.0 6.4 Table 2. Percentages of album parameter votes. Both iTunes and Media Player support the listed album parameters only on the song level. However, based on the results one should also consider implementing them on the album level. Both music players allow the users to rate single songs, while iTunes also has a parameter called album rating that is based on the average of album's individual song ratings. However, according to the results the participants would also like to rate entire albums by themselves. In the third part of the questionnaire, the participants were shown a list of 10 artist related parameters (Table 3). The results did not include any surprises and were inline with the previous findings. The most popular parameter was artist genre (85.1% of participants voted for A or B) and the least popular was the date when an artist was added to the music collection (46.8%). None of the listed artist related parameters are directly supported by the current versions of iTunes or Windows Media Player. However, based on the results one should also consider supporting this type of functionality on the artist level. Parameter A B C D Genre 19.1 66.0 14.9 0.0 My own rating 21.3 53.2 25.5 0.0 Automatic rating for an 6.4 61.7 25.5 6.4 artist (based on the average of artist's song ratings) Play count 14.9 63.8 17.0 4.3 Skip count 17.0 42.5 27.7 12.8 Date last played 2.1 53.2 38.3 6.4 Date added 8.5 38.3 44.7 8.5 Table 3. Percentages of artist parameter votes. Next, we studied how important genre related parameters were to the participants (Table 4). Compared to song, album, and artist parameters, there was a clear decrease in the popularity. In all cases except one ("my own rating for a genre"), the majority of participants answered that they would never use the listed parameters in their fictional music player. However, the formulation of questionnaire may also have affected the results. For example, in Table 8 most participants answered that they could use parameter "most played genres from the entire collection" either often or sometimes. Again, none of the listed parameters are directly supported by iTunes or Windows Media Player. Parameter A B C D My own rating 19.2 34.0 46.8 0.0 Automatic rating for a genre 10.6 34.1 53.2 2.1 (based on the average of genre's song ratings) Play count 4.3 40.4 51.1 4.2 Skip count 4.3 23.4 61.7 10.6 Date last played 0.0 36.2 51.0 12.8 Date added 0.0 17.0 72.4 10.6 Table 4. Percentages of genre parameter votes.
Page 3 ï~~Based on the results, the most interesting parameters for our future music visualization questionnaires are the popularities (play/skip count and ratings) of individual songs, albums, (genres), and artists. The visualization of tempo and release year has already been studied in  and , and the results of the latter should be usable also in the case of "date last played" and "date added" parameters. We have also studied the visualization of musical genres in two yet unpublished papers. 3.2. Song, album, artist, and genre collections The fourth part of the questionnaire consisted of 36 features related to collections of songs, artists, albums, and genres. In practice, these features are related to playlists and especially the generation of smart playlists. However, in the case of both iTunes and Windows Media Player, many of the proposed collections are either too tedious or impossible to create. According to the results, the majority of participants were interested in using every listed song collection feature either often or sometimes (Table 5). This implies that these features have deserved their place in existing music players. The most popular features were "most played songs from the entire collection" (91.6% voted for A or B) and "songs that have not been played yet" (87.2%), while the least popular were "recently played songs form a certain album" (55.3%) and "recently played songs from a certain genre" (59.6%). Feature A B C D Songs that I hate/dislike 27.7 38.3 31.9 2.1 Most played songs from the 42.6 49.0 6.4 2.1 entire collection Most played songs from a 23.4 53.2 19.1 4.3 certain genre Most played songs from a 38.3 46.8 10.6 4.3 certain artist Most played songs from a 19.2 57.4 21.3 2.1 certain album Least played songs 23.4 46.8 27.7 2.1 Songs that have not been 46.8 40.4 10.7 2.1 played yet Recently played songs from 34.0 44.7 17.0 4.3 the entire collection Recently played songs from 21.3 38.3 36.2 4.2 a certain genre Recently played songs from 10.6 66.0 17.0 6.4 a certain artist Recently played songs from 6.4 48.9 34.1 10.6 a certain album Recently added songs 44.7 40.4 14.9 0.0 Table 5. Percentages of song collection votes. In the case of album collections, most participants were interested in using every feature except one either often or sometimes (Table 6). The only exception was "albums that I hate/dislike", in which case only 48.9% voted for A or B. The most popular features were "most played albums from the entire collection" (93.6%) and "albums that have not been played yet" (89.4%). Feature A B C D Albums that I hate/dislike 21.3 27.7 48.9 2.1 Most played albums from 31.9 61.7 6.4 0.0 the entire collection Most played albums from a 19.1 55.3 21.3 4.3 certain genre Most played albums from a 12.8 55.3 23.4 8.5 certain artist Least played albums 12.8 53.2 29.8 4.2 Albums that have not been 40.4 48.9 10.7 0.0 played yet Recently played albums 14.9 55.3 25.5 4.3 from the entire collection Recently played albums 6.4 38.3 44.7 10.6 from a certain genre Recently played albums 8.5 53.2 27.7 10.6 from a certain artist Recently added albums 36.2 51.0 12.8 0.0 Table 6. Percentages of album collection votes. Without the exception of "recently played artists from a certain genre" (44.7% of participants voted for A or B), the majority of participants were interested in using every artist collection related feature either often or sometimes (Table 7). The most popular ones were "most played artists from the entire collection" (91.5% voted for A or B), "artists that have not been played yet" (87.2%), and "recently added artists" (87.2%). The least popular features included "recently played artists from a certain genre", "albums that I hate/dislike" (55.3%), and "least played artists" (55.3%). Feature A B C D Artists that I hate/dislike 25.5 29.8 44.7 0.0 Most played artists from the 34.0 57.5 8.5 0.0 entire collection Most played artists from a 19.1 53.2 23.4 4.3 certain genre Least played artists 14.9 40.4 40.4 4.3 Artists that have not been 40.4 46.8 8.5 4.3 played yet Recently played artists from 23.4 55.3 19.2 2.1 the entire collection Recently played artists from 6.4 38.3 44.7 10.6 a certain genre Recently added artists 27.7 59.6 8.5 4.2 Table 7. Percentages of artist collection votes. Feature A B C D Genres that I hate/dislike 14.9 25.5 59.6 0.0 Most played genres from 19.2 44.7 34.0 2.1 the entire collection Least played genres 4.3 21.3 72.3 2.1 Genres that have not been 10.6 21.3 66.0 2.1 played yet Recently played genres 10.6 27.7 55.3 6.4 from the entire collection Recently added genres 8.5 23.4 57.5 10.6 Table 8. Percentages of genre collection votes. In the case of genre collections (Table 8), the results were in line with the parameters for individual genres
Page 4 ï~~(Table 4). The majority of participants were not interested in unpopular, not played, recently played, or recently added genres. The only feature that received positive feedback was "most played genres", in which case 63.8% of participants voted for either A or B. Based on the results, the most interesting collection features for our forthcoming music visualization studies are most played, not yet played, and recently added songs, albums, and artists. The visualizations could e.g. include shortcuts to these collections, or the corresponding songs, albums, and artists could be marked with a certain color or symbol. 3.3. Selecting songs from a music player In the last part of the questionnaire, the participants were shown a list of different ways to select music from a computer or mobile phone, and asked how often they used each method. The answer options were: A = Always B = Often C = Sometimes D = Rarely E = Never F = No opinion In the case of computer, the most popular way to select music was to browse the list of artists (Table 9). 80.9% of participants used this method either always or often. Quite many people also selected their music by browsing the list of albums or songs. In the case of random/shuffle play, the votes were evenly divided between options B, C, D, and E. Method A B C D E F List of 17.0 63.8 12.8 4.3 2.1 0.0 artists List of 8.5 36.2 29.8 17.0 8.5 0.0 albums List of 10.6 31.9 19.2 25.5 12.8 0.0 songs List of 2.1 8.5 17.0 23.4 49.0 0.0 genres Random / 4.3 25.5 23.4 25.5 21.3 0.0 shuffle play Manual 8.5 8.5 27.7 38.3 17.0 0.0 playlist creation Automatic 0.0 4.3 21.3 25.5 46.8 2.1 smart playlists Table 9. Percentages of votes for selecting music with a computer. The least popular ways to select music included browsing the list of genres (72.4% did it rarely or never), automatically generated smart playlists (72.3% rarely or never), and manual playlist generation (55.3% rarely or never). While the unpopularity of genres was little surprising to us, the unpopularity of playlists can largely be explained by that fact that they are currently too tedious to create. It would thus be interesting to study how much their popularity would increase if new playlists could be created by simply clicking on some part of a visualization. In the case of a mobile phone, the results resembled those of computer and the popularity order of different methods was also the same. However, in many cases the average score was slightly less (i.e. closer to E) than in the case of computer. One potential explanation for this may be that the participants listened to more music with their computers than with their mobile phones. 4. CONCLUSIONS AND FUTURE WORK We presented the results of our online questionnaire that focused on the popularity of 71 selected music player parameters and features. The results are dominated by the Finnish participants. The majority of participants were interested in using most of selected song, album, and artist related parameters and features either often or sometimes. Still, the current versions of both iTunes and Windows Media Player support directly only a subset of them. Genre related parameters and features were clearly less popular than the others. The most popular way to select music from a computer or mobile phone was to browse the list of artists. The least popular ways included browsing the list of genres and generating playlists. In the future, it would be interesting to study how the popularity of playlists would increase if new smart playlists could be created by simply clicking on some part of a visualization. Based on the results, the most interesting parameters and features for our forthcoming music visualization studies are the popularities (play/skip count and ratings) of individual songs, albums, (genres), and artists, as well as collections of most played, not yet played, and recently added songs, albums, and artists. Other potential topics include, for example, symbols, facial expressions, friends, recommendations, and user's listening history. The results of all questionnaires will be used to design new user interfaces for music players, which will be then implemented and user tested. 5. REFERENCES  Apple iTunes 7.6, http://www.apple.com/itunes  Holm, J. Aaltonen, A. "Associating graphical objects with musical tempo", Proceedings of Audio Mostly Conference, Ilmenau, Germany, 2007.  Holm, J., Aaltonen, A. "Associating graphical objects with release year of music", Proceedings of IASTED-HCI Conference, Innsbruck, Austria, 2008.  Windows Media Player 11, http://www.microsoft.com/windows/windowsm edia/player/11/default.aspx