TIYU

A Location based Music Player for Sports

Georg Schneider and Henning Voss

Computer Science Department, University of Applied Science Trier, Trier, Germany

Keywords: Mobile Application, Web Geographical Information, Personalized Web Sites and Services, Multimedia and

User interfaces.

Abstract: This paper describes a mobile location based music player for outdoor sports like running, cycling or hiking.

Athletes can boost their performance while listening to the right songs. The TIYU system augments trails

with appropriate music for the athletes, either automatically using an intelligent selection mechanism or

manually via a web based authoring tool. These so created trails can be shared with other users on a web

based sports community platform.

1 INTRODUCTION

Many activities seem to be easier when we listen to

our favorite music. This is true for sports as well.

Findings from medical research have lately proved

the positive effect of music on the performance of

athletes (i.e. Karageorghis and Terry 1997,

Karageorghis et al. 2008). One consequence of these

findings is that portable music players have been

banned from official races in the USA 2007 (New

York Times 2007). On the other hand this is the

reason why the 2009 London Half Marathon (Run to

the Beat 2009) consequently adjoined music to the

race for everybody. At this event, bands were

playing along the racetrack. However it is not

always obvious to choose the right tune for a

workout. Some runners are faster than others and

when it comes to inclination or descents the pace

also changes (the same is true for other sports like

cycling or hiking equivalently). One way to deal

with these issues is that a runner would plan his

workout at home. Ideally she would select the trail

she wants to run on a map and select the appropriate

songs for the different sections of the run depending

on the profile of the trail. Using a mobile music

player with GPS she can automatically play the

before selected music according to her position on

the trail. The device could easily detect a pause in

the running and would stop the music accordingly.

However another approach would be that she could

simply run a trail and choose the songs manually.

Once she has finished her workout she could save

the settings for the future.

An even smarter solution would be that the mobile

device automatically selects appropriate songs

depending on the speed of a runner or alternatively

in a way that she would accomplish to attain a

certain training goal, e.g. run a certain velocity.

Once the different playlists for the trails are set up

the runner possibly wants to exchange either

playlists or trails with other runners or even meet

other athletes with similar preferences. She would

possibly be interested in what music other people

listen to at the same trail or if there are runners that

have similar characteristics concerning velocity,

duration of the workout, time of the day of the

training etc. like her.

Another idea could be to see what playlist would be

appropriate in order to improve the running time for

a certain trail.

In order to realize such an application we will

describe in this paper a mobile application together

with a web based community website where users

can exchange the type of information mentioned

above. The name TIYU is derived from the Chinese

language (体育) and means “sports“ or “making

sport”.

First we will describe systems with similar goals in

the related work section. Afterwards we will

illustrate the concepts used in order to build such a

system. Then the concept is refined in the realization

section. We will show an example of the system and

118

Schneider G. and Voss H.

TIYU - A Location based Music Player for Sports.

DOI: 10.5220/0002807201180124

In Proceedings of the 6th International Conference on Web Information Systems and Technology (WEBIST 2010), page

ISBN: 978-989-674-025-2

give a short overview of the implementation. We

will finish the paper with a conclusion.

2 RELATED WORK

Several approaches to combine music and sports

exist already. The MiCoach system (MiCoach 2009)

from the companies Adidas and Samsung uses

sensors for heart rate and pace, which are

transmitted to a mobile phone. The phone uses this

information to select appropriate songs once the user

has selected the sport she wants to perform, e.g.

running or cycling and the desired speed. The user

has to copy songs to her mobile phone and set up a

playlist, which she listens to during her workout.

MiCoach offers so called motivational songs, which

can be selected manually during the workout. Songs

can also be selected automatically depending on the

type of sports the user has selected. A GPS sensor is

not a part of the system. A belonging website is used

to personalize the system, to set up training plans

and to visualize data acquired during the workout.

Nike+ (Nike 2009) is a cooperation between the

companies Nike and Apple. It consists of a mobile

music player, the iPod and a sensor that counts the

steps of a user. The user can define playlists or

download playlists that are supportive for a certain

type of sports e.g. running or cycling. A community

website is used to define workout goals and to

visualize and exchange information concerning the

training, like consumed calories. Furthermore

running trails can be defined and shared on the

website using a map. Since the iPod has no GPS

receiver these data cannot be used directly for a

workout.

The Nokia Sports Tracker (Nokia 2009) is an

application that runs on Nokia cell phones with GPS

receiver. Sportsmen can save GPS data of their

trails. These data can be transmitted to a website and

visualized on a map. Additionally this information

can be shared with other users. The so called “Live

Sharing” functionality transmits automatically the

data to the web server and other users can see the

location of this specific user. In the case of the

Nokia Sports Tracker, there is no support integrated

for music as part of the training.

Active Outdoor (Wayfinder 2009) is a website

where users can upload GPS tracks in order to

visualize these data on a map. The system can be

used for navigation and sharing of the GPS tracks

with other users. The system mainly supports blogs,

bulletin boards and the creation of groups. Similar as

the Nokia Sports Tracker, there is no support for

music.

Especially for runners more specialized websites

and communities have emerged lately, like Jogmap.

In general they differ only slightly compared to the

two foregoing systems concerning the functionality

in regard to storing, visualizing and sharing GPS

information. Jogmap offers a greater variety of

possibilities in order to create personalized workout

plans and analyze data for runners.

In contrast to the systems and services illustrated

above the service Run2rythm (Blake 2009) produces

music especially targeted for running. Users of the

service can select tunes based on the velocity they

want to run. Figure 1 illustrates the connection

between BPM (beats per minute) of the music and

time and distance that a runner will cover running in

the beat of the music.

Figure 1: Dependency between music and running pace

(Run3rythm 2009).

However this service does not include functionalities

like tracking the trail a user runs.

None of the systems above offers functionalities like

we have motivated in the first section. Either a GPS

device is integrated in order to visualize and track

the trails or music is used to support a training goal.

Using both, music and GPS data together to annotate

trails, to automatically select suited songs and to

share this information among users of a web

community has not been realized so far.

In the following we will describe the ideas of the

TIYU system, which combines the different

approaches of music selection in dependency with

the kind of sports and velocity of a user together

with the trail she moves along. A mapping and

visualization of this information using GPS and

maps is also part of the system.

TIYU - A Location based Music Player for Sports

119

3 CONCEPTS OF THE TIYU

SYSTEM

The goal of the TIYU system is to integrate the

different ideas of music selection, mapping and

exchanging of GPS data and sharing this information

between users into one system. First of all we will

describe different ideas of music selection for

outdoor sports.

3.1 Music Selection

In the preceding section we have already illustrated

the different influences in regard to music selection

for sports. Moving along an outdoor trail, not only

the kind of sports and the targeted speed have an

impact in the music selection but also the profile of a

specific route, especially the altitude profile. If a

user moves along a river the trails is normally flat

and the speed is relatively constant as well.

Compared to a trail in the mountains the situation is

different. The athlete will most certainly move

slower uphill than downhill.

Additionally the situation may arise where the user

wants to listen to a certain tune at a specific section

of the track but the tune may be to short or too long.

In the following we will sketch our ideas to select

music for a certain outdoor trail. In the following,

we will refer to a playlist, where the songs are

related to a certain GPS position as Location

Playlist. We will concentrate in our demonstration

on running but the same ideas apply for other sports

equivalently:

 Manually, on the fly

While running, the user selects manually one

song after another. The playlist is recorded

together with the GPS data as a location

playlist.

 Manually, prepared earlier

A user has already set up a location playlist

for her trail. This may be the case either using

one of the playlists created using the former

procedure or choosing a playlist, which has

been created by another person.

 Automatically, user centered

Automatic selection of songs can be achieved

using the findings illustrated in figure 1. Using

GPS, the mobile device can compute the

velocity of a user and likewise can choose an

appropriate song for the current situation.

 Automatically, trail centered

If the trail is known in advance or if the

mobile device uses the GPS data in order to

find out the profile of a trail, the system can

figure out, which song might be appropriate

for the current situation. However this mode

strongly benefits from a tagged music library.

These tags may be specified manually or

calculated after a run (e.g. songs that a user or

different users often play while running

uphill)

As we mentioned earlier situations may arise, where

a song is too short or too long for a certain section of

the trail. In the case, where songs are too long, they

may be skipped or blended into the next song. For

songs that are too short the situation is more

complicated. Different approaches may be possible:

 The song is simply repeated until to the next

position, which is identified with a different

song.

 Sections of a trail are not identified with a

single song but with a group of different

songs. In this case songs from this group are

played until the user reaches the next section

of the trail.

 The song is artificially slowed down or sped up

in regard to the current speed of the user (e.g.

Bieber and Diener 2005).

3.2 Audio Feedback

Audio feedback is a way to give information about

certain aspects of a workout or trail to a user. In the

TIYU system we regard aspects of the workout e.g.

if a users moves too fast or too slow or if she moves

away from the selected GPS trail. In order to inform

the user about these facts we do not want to interrupt

the music displaying a voice message. This could

bother the user and interrupt the rhythm of the

music.

Our approach is to inform the user by turning the

music up or down. Thus the user can quickly

identify that she is no more within the intended

frame of the workout and she can check the trail or

the speed.

3.3 Workout Information

In order to share workout information between users

and possibly find training partners information like

day, time of the day, sport (e.g. running, cycling),

distance, duration and favorite trail and music are

stored in the system as well. This information can be

used to find users with the same training behavior.

This information is crucial for a community with

numerous users because it enables searching and

filtering of relevant potential training partners or

suited trails or location playlists.

WEBIST 2010 - 6th International Conference on Web Information Systems and Technologies

120

4 REALIZATION

In the following sections we will describe the

realization of the system. We will start with

illustrating our approach of the calculation of an

automatic song selection process.

Afterwards we will illustrate the procedure of

selecting users with a similar training profile.

Finally we will describe the architecture of our

system.

4.1 Calculation of Suitable Songs

Selecting the right song for the actual setting of an

athlete is crucial for our system. In order to make an

automatic selection for the best song that can be

played in a specific situation we use the following

approach:

BEGIN

bpmBest = 2.22 * geschwindigkeit + 140

bpmDiff = bpmBest – bpm(lied)

IF bpmDiff < 0 THEN

bpmDiff = bpmDiff * -1

bpmPoints = 255 - 1.0 / 10.0 *

bpmDiff * 255

IF bpmPoints < 0 THEN

bpmPoints = 0

RETURN bpmPoints

END

First we calculate the optimal beats per minute

(bpmBest) for the current velocity using the findings

illustrated in figure 1.

Afterwards we calculate the bpmPoints, which has a

highest value of 255. If the beats per minute do not

fit exactly to the speed of the user, the bpmPoints for

this song will be lower, e.g. if the bpmDiff is 5, the

value will be 127. If the difference bpmDiff is 10

the value is 0. Basically the function shows a linear

decline. The different values are based on

experiences from several tests we have carried out.

4.2 Display of Training Partners

The system supports the user in order to find suitable

training partners or location playlists. Thus not only

the music tastes of the different users but also their

training schedule and profile has to be taken into

account.

Because it concerns finding users with the same

preferences for music the Lastfm platform (Lastfm

2009) already offers a web service to find out similar

music given a certain music profile.

Additionally, the training schedule and profile

have to be taken into account. We have used the

following formula in order to calculate the similarity

of users:

Similarity = SameSports/4 + SameDay/4 +

SameTime/4+ TrainingIntensity/4

(1)

SameDay = SameDayMonday +

SameDayTuesday+…

(2)

SameDayMonday =

min(SameDayMondayUser1/TotalUser1,

SameDayMondayUser2/TotalUser2 …)

(3)

In (1) we calculate the overall similarity between

users, summing up the different similarities of the

factors we take into account. Formula (2) shows the

calculation of the similarity concerning the days

different users workout, whereas for each day (3) the

different users are compared in regard to their

training behavior on that day in relation to their

overall training.

The remaining similarities are calculated

accordingly.

4.3 System Architecture

The architecture of the TIYU system is a client

server architecture as illustrated in Figure 2.

The mobile user (User1) communicates with a

web server after a training or even during a training,

when a mobile connection is possible and allowed

by the user. Information about GPS position, playlist

and further application relevant information (e.g. if

the song has been played on an incline or a decline)

is transmitted to the server using the HTTP protocol

for communication and the XML format for the

content description.

Later the user can visualize or edit the data using

a web interface (e.g. User2).

Other users may download already existing

location playlist form the website, for example

User3 and start the training using these data.

User4 may set up her training using a web interface

before the training and associate songs or groups of

songs to certain sections of the trail. Later on she

uses the location Playlist for her workout (e.g.

User5).

TIYU - A Location based Music Player for Sports

121

Figure 2: Architecture of the TIYU system.

Figure 3: TIYU website.

5 EXAMPLE

The following example shows different views of the

system. The TIYU website as shown in figure 3

consists mainly of a map displaying the different

trails the user has (or friends of the user have)

created in this belonging section. By clicking on the

different numbers on the map, information about

length of the trail, velocity and time the trail has

been used for the last time will be displayed.

Furthermore the system tries to detect automatically

the kind of sports, e.g. running, hiking or cycling

depending on the velocity of the user.

In a different view the users of TIYU who

upload continuously their GPS coordinates to the

WEBIST 2010 - 6th International Conference on Web Information Systems and Technologies

122

server (and who allow to access these data) will be

visualized similarly on the map.

The website can also be used to exchange

comments to a specific trail or visualize training

data. The user can see on which days she usually

works out. Furthermore she can see the total amount

of time and distance she has covered in this week

and she can also look at the altitude profile for a

certain trail and the belonging velocity. Finally she

can search for similar users concerning training

behavior and music taste.

The user interface on the mobile device can

cover only a very limited space. Figure 4 shows one

view of the mobile client running on the cell phone.

In this case the training data is displayed showing

the velocity for a certain trail in the first minute.

Further views can display the dependency of

distance and velocity. Another graph display shows

the altitude profile together with time an athlete has

covered. Altitude profile and distance is a further

alternative.

Additionally the current song and a table with

statistics for the training so far are possible screens.

A map view showing the position of the user and

also the selected trail if desired, can be displayed on

the mobile phone as well if a mobile data connection

is available.

Figure 4: Screenshot from the mobile client.

6 IMPLEMENTATION

The TIYU system has been implemented using

JavaMe on the Nokia cell phone E90 Communicator

(MIDP 2.0) with 128 MByte memory, a 330 MHz TI

OMAP2420 processor (ARM11 architecture) and a

Symbian S60 operating system. The songs are

played using the media player installed on the

phone. In the current version of the system music

streaming is not yet available and the songs have to

be copied on the phone.

The server application is realized in Java as Servlets

hosted on a Tomcat web server. As database

MySQL is used together with the Hibernate

framework. The web application makes use of the

Google Web Toolkit. Furthermore the Google Maps

web service is used in order to display the map data.

The Lastfm web service provides the data for the

similarities of different songs.

7 CONCLUSIONS

In this paper we have presented a concept for

athletes who are interested in working out with

music. The system offers different modalities.

A user has different possibilities to generate a

location playlist, like generating one using a web

application, generating a location playlist on the fly,

downloading a playlist or let the system decide

automatically.

Furthermore information of the trainings are

recorded and displayed on a community website in

order to exchange trails or location playlists. The

discovery of new music, new trails and possibly also

future training partners can also be stimulated using

the system.

We have made a first test with three different

users that are regular runners. The feedback was in

general positive. The users characterized the system

as beneficial. Although these opinions are promising

a more detailed and profound user study has to be

carried through in the future in order to gain a valid

insight in the usefulness and the behavior of the

system.

REFERENCES

Bieber, G., Diener, H.; StepMan - A New Kind of Music

Interaction, in: HCI International 2005, Lawrence

Erlbaum Associates, Inc., Mahwah, New Jersey, 2005

Blake, G.; Run2Rhythm Productions Pty Ltd, How

Run2Rhythm started, http://www.run2r.com/about-

run2rhythm+about-run2rhyth-music.aspx, [4.11.2009]

Jogmap media GmbH, http://www.jogmap.de/ [4.11.2009]

Karageorghis, C., Jones, L., Stuart, D.; Effects of Music

Tempi on Music Preference, Intrinsic Motivation, and

Flow during Long-duration Exercise, International

Journal of Sports Medicine, Georg Thieme Verlag

2008.

TIYU - A Location based Music Player for Sports

123

Karageorghis, C., Terry, P; The psychophysical effects of

music, in: Sport and exercise, United States Sports

Academy, 1997

Last.FM Ltd., API Spezifikation, tasteometer.compare

Web Service, http://www.lastfm.de/api/

show?service=258, [4.11.2009]

MiCoach, www.adidas.com/miCoach/ [4.11.2009]

New York Times, 2007, http://www.nytimes.com/

2007/11/01/sports/othersports/01marathon.html,

[4.11.2009]

Nike, Inc., Nike+: Hear How You Run,

http://www.nike.com/nikeplus, [4.11.2009]

Nokia Oyi, Nokia Sports Tracker,

http://sportstracker.nokia.com/nts/main/index.do,

[4.11.2009]

Run2Rythm, http://www.run2r.com/Technical+linking-

bpm-to-running-speed.aspx, [4.11.2009]

Run to The Beat, http://runtothebeat.co.uk/ [4.11.2009]

Wayfinder Systems, Inc., Active Outdoor,

http://www.activeoutdoor.com/Activeoutdoor/loginfor

m.js [4.11.2009]

WEBIST 2010 - 6th International Conference on Web Information Systems and Technologies

124