Mapping Detected Periodic Dance Movements to Control Tempo in the Music Playback of Electronic Dance Music

DEGREE PROJECT IN THE FIELD OF TECHNOLOGY MEDIA TECHNOLOGY AND THE MAIN FIELD OF STUDY COMPUTER SCIENCE AND ENGINEERING, SECOND CYCLE, 30 CREDITS STOCKHOLM, SWEDEN 2019

Mapping detected periodic dance movements to control tempo in the music playback of Electronic Dance Music

LILIAN JAP

KTH ROYAL INSTITUTE OF TECHNOLOGY SCHOOL OF ELECTRICAL ENGINEERING AND COMPUTER SCIENCE

ENGLISH - ABSTRACT Mapping detected periodic dance movements to control tempo in the music playback of Electronic Dance Music

Engaging in the music set of one’s favorite artist or DJ is oftentimes leading to the result of a powerful andeuphoric felt experience, a sensation partly also induced from dancing in beat to the music. In an attempt to simulate a similar dance experience, a user-study was designed in order to investigate when a user is let to dance in rhythm to a music playback and in addition, in control of a music playback tempo through theinduced dance movements. A proof-of-concept prototype was built and tested in an initial study, followed by a main study where the prototype had been modified and 12 participants participated. A questionnaire was given containing various question statements to be rated through a Likert-scale regarding their subjective experience. Open-ended questions were also included to collect their own opinions. From the results, an enhanced engagement and enjoyment of the music could be identified when being able to manipulate the tempo.

SVENSKA - SAMMANFATTNING Kartläggande av analyserade periodiska dansrörelser för att kontrollera tempo av elektronisk dansmusik Att engagera sig i ett musik-set av ens favoritartist eller DJ leder ofta till resultatet av en kraftfull och euforisk känsloupplevelse, en känsla delvis framkallat av att man dansar i takt med musiken. I ett försök att simulera en liknande dansupplevelse undersöker denna användarstudie när en användare dansar i rytm till musik och dessutom är i kontroll av tempot genom de skapade dansrörelserna. En proof-of-concept prototyp konstruerades och testades i en första studie, följt av en huvudstudie där prototypen hade modifierats och 12 deltagare deltog. Ett frågeformulär gavs med olika frågor som skulle bedömas via en Likert-skala, med avseende på deras subjektiva erfarenhet. Öppna frågor ingick också för att samla deras egna åsikter. Från resultaten kunde ett ökat engagemang och en förhöjd njutning av musiken identifieras när man kunde manipulera tempot. Mapping detected periodic dance movements to control tempo in the music playback of Electronic Dance Music Lilian Jap KTH Royal Institute of Technology Stockholm, Sweden [email protected]

ABSTRACT [8]. Current evolving musical/technical landscapes has shifted the focus away from the passive individual Engaging in the music set of one’s favorite artist or DJ towards an active role into sound [30] with the impact is oftentimes leading to the result of a powerful and of embodied interaction. As a result, it might exist a euphoric felt experience, a sensation partly also need to explore what possibility the processing of our induced from dancing in beat to the music. In an movements may possess in the context of EDM. attempt to simulate a similar dance experience, this user-study examines when a user is let to dance in If the music were to be influenced by movements, a rhythm to a music playback and in addition, in control different dynamic flow is probable to appear in the of a music playback tempo through the generated interaction. Being able to find an organic way for the dance movements. A proof-of-concept prototype was user to independently drive the decisions of how to built and tested in an initial study, followed by a main control the sound parameters of a music playback, this study where the prototype had been modified and 12 could pave the way for a more inclusive musical participants participated. A questionnaire was given environment. The indicated might open up a field for containing various question statements to be rated individuals wishing to physically interact with music through a Likert-scale regarding their subjective on a personal level – without the requirement of having experience. Open-ended questions were also included a musical background nor the expectation of having the to collect their own opinions. From the results, an actual physical ability to play current existing enhanced engagement and enjoyment of the music instruments. could be identified when being able to manipulate the tempo. Along with possibly gaining an equivalent positive effect sparked from dancing, it should also be Author Keywords encouraged to maintain the bridge between embodied Interactive dance; IMU; real-time; sonic interaction design; user- interaction and interfaces involving musical study; Max/MSP; yin; expression. Not the least within interfaces making use of real-time interaction, as novel forms of it are still 1. INTRODUCTION being sought for e.g. educational contexts, artistic or entertainment purposes, physical rehabilitation and In Electronic Dance Music (EDM), a DJ combines in- health, etc. advance planning and real-time decisions for the purpose of hopefully creating an intense and ecstatic In this thesis, the following research question is dance experience. Taking advantage of the same felt intended to be answered; What mapping methods of experience could be a design aspect worth considering real-time measurements of rhythmic movements of a when finding new practices for us as users interacting dancer’s body to tempo manipulations in EDM can with a system. Through bodily interaction with a lead to a dance experience that compares positively to system, we are able for example to create, manipulate a standard playback of the same type of music? For this or communicate meaning. This is an understanding that context, the definition of a dancer entails any possible provide the basis to the idea of embodied interaction user of the proposed system; whether the user possess

the skillsets of a dancer or not. A proof-of-concept a way towards implementing dynamic human- prototype is presented, its design process and findings computer interaction (HCI) designs further. As it is a made after performing a user-study where participants way of handling gestures to not primarily be got to interact with the prototype. considered as goal-directed tasks, instead gestures can deliver a qualitative meaning to the interaction [34].

2. BACKGROUND 2.2 Various interdisciplinary approaches The following section provides an outline of topics The aforementioned is stated to be limited to the related to gesture-based sonic interaction. Firstly, context of musical performance. However, motion musical gesture as a concept is explained. Secondly, a sensors used with the objective of implementing closer review of various applied fields of the concept is different sound synthesis techniques can likewise be presented as well as a motivation for why further seen being implemented in various mixed investigating novel forms of the concept is needed. interdisciplinary approaches. Lastly, an in-depth review of previously used methods and technologies relevant for this work is presented. Methodologies of different approach contribute to re- contextualize technology and its role. In the context of 2.1 Embodied music interaction various artistic practices, sonification through the use A number of studies involving experiments with music of sensors is suggested as a way of enhancing the and embodiment can be found. In the context of artistic purpose. In the work of Elblaus et al. [9] musical performances, the found work appears to have involving contemporary circus, the interactive sensor- a common feature; musical gestures as an extension of based motion sonification system resulted to be the body. Some characteristics of these musical effective in terms of composition, performance and gestures will inevitably be reflected in the resulting response from audience. While endorsing methods in musical tones when all sounds from traditional acoustic dance can successfully be used in circus arts to create instruments are produced by biological motion of expressive performances, the technology further humans [6][7]. Movements from musicians such as widened the possibilities for creating expressive head shakes or body sway are examples that - although performances in dance-related circus practice. not having an active role in the sound generation - still can serve a communicative purpose of their own [6][7]. Additionally, gesture-based interactions have also been used for a learning purpose as proved by Maes et al. Musical gestures as suggested by Jensenius et al. [15] [19]. In helping users understand how musical meter are “human body movement that goes along with can be articulated into body movements, their meter- sounding music … the gestures of those that produce mimicking gesture-based tool demonstrated a the sounds (the musicians), and the gestures of those relatively shorter required learning period along with a that perceive the sounds (the listeners or dancers)”. clearer understanding of the intended behaviour. For Rarely there are works about gestures of the latter. It is the same purpose, Gelineck & Böttcher [10] presented however no surprise that the exploration of novel an educational tool with the intention of letting novice musical interfaces has their base in the aforementioned, users – in this study being non-technical conservatory since they are often based on the relationship between musicians – be introduced to novel musical interactions the musician and the musical instrument. Gained provided by new existing technology. Making use of insights from these are in most cases originating embodiment as a central element in musical precisely from the study of musical gestures producing performance is likely to enhance the pedagogical sounds [34]. approach, as was the hypothesis of Bevilacqua et al. [1]. Visi et al. [34] aimed to explore issues related to the representation and mediation of body movement in a When placing wireless motion sensors on major parts musical performance through digital data collected of the body - in the case of Jung & Cha [16], the hands from wearable motion sensors. Using motion data in and feet – it was shown that utilizing gestures as game combination with traditional instruments in musical content bring more substance to the game. Users of a practice implied that extracting movement features are game are often in better control when utilizing the body

3 [16] and a larger enjoyment is more likely to be metrical hierarchy, each level studied being one-beat, achieved into the interaction, even if only tracking the two-beat and four-beat [31]. Different body parts also small body gestures such as hand drawing [14]. differ in terms of ability to synchronize [2]. Burger et al. [2] showed that the feet and hip were more In a musical performance, the interactive work could accurately synchronized to different beat levels than be explained as coming into existence between the user the hands and head when asking their subjects and the designed application system – the instrument - individually to dance freely yet synchronized to six when the two parties are both listening and performing Motown songs of different core tempi (105, 115 & 130 [5]. This is utilized in this approach of designing a BP). On the other hand, repetitive movements are novel sonic interaction. Particularly when it comes to resulted to be more pronounced in hand gestures than hearing rhythm in music, bodily movements can play the gestures of other body parts [17]. an important role when developing the skill involved in rhythmic perception [5]. In this context, the present Based on the preceding, measuring the rhythmicity in work intends to further investigate the effects when the human body specifically in relation to EDM is moving periodic to music – by both only listening as proposed to be further investigated. In the work of well as listening and performing. Similar to how a Zhao et al. [37], an Inertial Measurement Unit (IMU)- chain of events are often repeated in music, the same signal is re-interpreted as if it was an audio signal and concept can be applied to the human body as an processed. Using a time-domain pitch tracking instrument producing rhythmicity. algorithm, repetitive patterns are predicted by detecting the “pitch” of the inertial gait measurements. For Audio feedback can induce more awareness as it brings example, the YIN-algorithm in which an deeper understanding in how the body moves [12][35]. autocorrelation-based pitch tracking method enables to Aligning rhythmic movements with rhythm in music rate the quality of the detected periodicity [4]. have been used as an approach to improve the user’s own movement performance in the context of sports The genre of EDM is perceived as being repetitive and (running) as well as physical rehabilitation and health static with a clear rhythmicity and continuous strong [12][13][18]. beat, even for the untrained ear [28]. In order to proceed with the suggested, examining various 2.3 Computational analysis of rhythm in audio and approaches on how to estimate and manipulate tempo movement of an audio stream in real-time are to be considered for Decomposing the issue of how a dancer interacts with possible implementation. One attempt in confronting music and space, Leman & Naveda [17] contribute to the task of tempo estimation was done by Griffin et al. the idea that musical cues are projected onto the [11]. The authors found an engaging way for the user repetitive gestures of professional dancers in Samba & to create mash-ups of music clips by combining phase Charleston. These music-driven repetitive gestures vocoding and beat-tracking technology, and in result generate dance patterns, which are induced from successfully synchronize the playback of two clips in musical cues. Naveda et al. [20] proposed a an autonomous way. However, the input of audio clips methodology to identify metrical accents in more in the authors’ work is to be done manually, as for “freely” movements – unconstrained movements – many commercial DJ-softwares (Serato, Virtual DJ, based on two features; level of accumulative velocity Traktor etc.). Oliveira et al. [22] proposed in their work and density of directional changes. The relationship INESC Porto Beat Tracking software (IBT), an audio between metrical structure of music and organization beat tracking external for Max/MSP, PD and available of continuous and discrete events in the movements, is for open-source. In a predictive manner, the beat- suggested to be described through the density of events tracking system identifies beats through causal and continuous display of energy. decisions over incoming audio input signal. Hockman et al. [13] used the approach of locating the frequency Similar is suggested from Toivainen et al. [31] where of events understood as footsteps through tempo focus lied on how pulsations on different musical estimation and in turn adapting the tempo of music metric levels manifest itself in music-induced playback in accordance to the pace of a runner. movements. Typical movements were identified in the Robertson & Plumbley [25][26] proposed B-keeper, a

real-time beat tracker used to synchronize tempo of 3.1 Initial study pre-recorded music to a live drummer. Today it also exists as an extension, BeatSeeker, which is adaptable An initial study was conducted on three participants to to any rhythmic real-time audio signal instead of a gather more detailed knowledge about the gained data manual triggering of drum signals. signals from the IMUs. For the participants to feel at ease and be able to express without any limitation, this 2.4 Optical Motion Capture vs IMU was performed without any recordings of sensor data. Previous works regarding sensor systems have been Annotations were however made of the participants’ studied and a number of advantages/disadvantages expressions. Focus was also put on examining different have been identified. The marker-based infra-red placements of the sensors on the body in terms of Motion-Capture (MoCap) technology is acknowledged usability as well as pronouncement in the movements. for measuring complex movement in a three- One sensor was used and based on the notion that the dimensional space. Being a few decades old it is placement should serve a practical fit, five different roughly recognized as being one of the main types of placements were tested; left & right hand wrist, right motion-tracking technology [29][33]. Due to its ankle joint, hip and lastly, the back. An example of one reported precise tracking as well temporal resolution, it placement can be seen in figure 1. allows to measure detailed finger movements and Investigating earlier implementations of processing facial expressions while performing real-time human sensor data in conjunction with audio signals and movement analysis [33]. However, the technology’s music playback in real-time were also included in the requirements limit itself to a defined space. The system initial study. This was done in order to find an encompasses constraints that can be crucial for the end appropriate method for the context of giving the user design, such as requirement for calibration before use, independent control and unlimited movement in an limitations in movements and difficult for public undefined space. Approaches suggested from spaces [21][29]. [4][13][17][20][22][23][25][26][27][31] were studied, experimented and examined. Outcomes obtained at this Most recently have the use of IMUs from various stage of the process were adopted in the used manufacturers risen up to the surface and emerged its prototype. way into movement-based interaction design [3][21][29][34]. Their miniature in size, mobile use and reported accurate essential values make them beneficial to use and tend to be more commonly utilized [35].

3. METHOD. For this study, a proof-of-concept prototype was designed in order to investigate periodicity in a user’s body movements. In section 3.1, the initial study regarding both the prototype’s apparatus and user Figure 1: Test placement of sensor during initial study. interaction is described. An overview of the prototype’s components is then outlined in section 3.2. 3.2 Prototype Final evaluation was later assessed by gathering 3.2.1 Equipment and platform qualitative data using methods described in section 3.3 for the purpose of measuring the subjective experience The technology used for this work will consider the when using the prototype. Next Generation of Inertial Measurement Units (NGIMU) from x-io Technologies Ltd1. In contrast to

the manufacturer’s former original IMU sensor, the NGIMUs makes use of the Open Sound Control (OSC) protocol. This opens up compatibility with other

1 http://x-io.co.uk/ngimu/

5 Figure 2: An overview of the chain of operations in prototype software applications, e.g. Max/MSP, Unity along with available on GitHub3. A screenshot of the prototype’s libraries for programming languages such as Python, Graphical User Interface (GUI) can also be seen in C++, Matlab. Accordingly, OSC was the figure A.1 in Appendix A. communication protocol used in this project. The chain of operations: The platform chosen to collect and process incoming data from the NGIMU sensors was Max/MSP 72. The 1. OSC messages about the accelerometer data from a real-time communication was performed via Wi-Fi NGIMU sensor are received. using TP-Link AC750 travel router as a separate 5Ghz 2. Raw accelerometer data from x-, y- and z-axis are wireless network instead of the sensors internal smoothed through low-pass filtering provided by antennae. This allows for multiple sensors to be Max/MSP [slide]-object, filtering logarithmically (See communicated [32]. Figure 3 displays an overview of eq. 1) with slide value 10. A given sample output this equipment. As the host computer, a mid-2014 13- from slide is equal to the last value (yn-1; n=sample inch MacBook Pro Retina was used. number) plus the difference between the input (xn) and the last value divided by the slide value. Given a slide value of 1, the output will therefore always equal the input. Given a slide value of 10, the output will only change 1/10th as quickly as the input4.

푥 − 푦 푦 = 푦 + ( 푛 푛−1 ) (eq. 1) 푛 푛−1 푠푙푖푑푒−푣푎푙푢푒

3. The fundamental frequency (F0) of the data stream in each axis is estimated by the [pipo.yin]-object from Mubu for Max-toolbox5. The object makes use of the

Figure 3: Equipment used. Upper left: Anker PowerPort 10, YIN-algorithm [4], which also provide the quality TP-Link AC750. Lower left: NGIMU body strap, two NGIMU factor of the detected periodicity. Several values were motion sensors, USB cables for charging and communication. tested for different attributes of the yin-object, and significant for the interaction were the sample size of 3.2.2 System design frame (N), hop size (N/16 [13]) & framerate (sample An overview of the system design can be seen in figure rate of sensors). N = 100 were determined to provide 2. Parts of the operations were computed using enough accurate results while also being short enough JavaScript, within the Max/MSP environment. Values to track a speed up/speed down of the acceleration. mentioned in the following operations were decided on With the sensor’s sample send rate of 50 Hz, this will after exploring different values in the initial study, for create a buffer size of 50 frames of 100 samples each. the purpose of providing a satisfying real-time A more detailed description of this step can be found response with reduced delay. Full patch has been made in figure A.2 in Appendix A.

2 https://cycling74.com/products/max/ 4 https://docs.cycling74.com/max7/maxobject/slide 3 https://github.com/lilianjap/periodicity_detection 5 http://forumnet.ircam.fr/product/mubu-en/ 6

4. Values of the detected quality factor – in each axis – occurred. Figure 5 shows screenshots from two are given and analysed using two moving averages participants’ recordings. with subsets of 10 resp. 5 sample values. The participants were asked if wanting to choose music 5. Detected highest quality factor determines which themselves - if not, the songs were chosen randomly axis’ estimated F0 to be used. within the range of 110-140 BPM and within the EDM- 6. F is converted into beats per minute (bpm). genre (see Appendix B for a complete list of used 0 songs). The genre displays repetitive rhythmic and 7. Based on the changes in speed of the movements, the melodic features [29] and can be considered as one of bpm will change accordingly. When detecting a speed today’s popular music genres [36], thus a genre up/down, the current bpm will change by with a step familiar to participants and therefore greater ability to increase/decrease of 3 bpm. induce movement. Each session contained different songs and kept within 15 minutes as a way to keep the 8. Bpm values are sent to a sample-playback object to participant engaged. be time-stretched in real-time 67 . The time-stretching makes use of beat annotations of the played audio sample (see block 9 in figure 2). 3.3 Main study The goal of the main study was to evaluate the subjective experience when using the prototype. This was carried out by performing experiments comparing the participants’ dance experiences without and with the prototype.

3.3.1 Participants 12 participants between the ages 22-31 participated in Figure 4: Environment setting of the experimental setup. the study (8 men and 4 women, mean age 27 years). 9 3.3.3 Questionnaires participants have a background in dance or are working professionally with dance. All participants were After each session, the participants were given a reported to be in a healthy condition. Each participant questionnaire to fill out (see Appendix C). These was recorded individually and approved consent before entailed both open- and closed-ended questions to the experiment started. gather qualitative resp. quantitative data. First part of the questionnaire focused on estimating the 3.3.2 Experimental setup participant’s experience to standard playback of the Two experiments were performed for each participant. music. Second part of the questionnaire focused on The first experiment consisted of the participant given estimating their experience in relation to the previous the task to move freely, but repetitive to the presented session, but also to collect their own thoughts and music stimuli. The second experiment included the opinions regarding the interaction that is involved in same task but the ability to control the tempo of the the second experiment. music stimuli. The NGIMU was in both sessions placed on the right hand wrist of the participant. The experiments were conducted in a personal living room using 2.0 stereo speakers with Bluetooth for playback of music (figure 4). The participants were also video recorded to collect observations of their spontaneous reactions as well as what kind of interactions that

6 https://docs.cycling74.com/max5/refpages/max-ref/transport.html 6 https://docs.cycling74.com/max5/refpages/msp-ref/groove~.html

7 than before. However, a minimum frequency was set to 1Hz giving a lower tempo threshold of 60bpm. 4.2 Main study In all the sessions, none of the participants chose to choose music themselves. The music stimuli were Figure 5: Two of the participants testing the prototype in the therefore randomly chosen for each session. Right after main study. each session was finished, each participant got to fill 4. RESULTS out a questionnaire containing various question statements to be rated on a 5-point Likert scale as well This section will present data gathered from the initial as open-ended questions to gather their own thoughts study and the main study. about the user-test (See Appendix C for a complete list of questions). 4.1 Initial study 4.2.1 Engagement in music Three participants took part in this stage of the study. Expressed preferences from the three were to place the Figure 6 shows the result when asked about how sensor on either their ankle joint or hand wrist. One of engaged in the music they were, from both first and the participants explained placing the sensor on the second session. The x-axis represents the distribution ankle joint made it less noticeable and could therefore of the participants rating, where 1 = not at all engaged, focus on the dancing more. While another participant 2 = little engaged, 3 = moderately engaged, 4 = very preferred to place it on the hand wrist because it engaged and 5 = completely engaged. The y-axis enabled more control when manipulating the tempo. represents the number of participants. First session without tempo manipulation gave mean value rating of When investigating how the sensor data would display 3.42 while the second session with the ability to in the participants’ movements, the largest manipulate tempo gave a mean value rating of 4.00. pronunciation of the movements could be found when placing the sensor on their hand wrists whereas the ankle joint would display a moderate pronunciation the movements, the hip and the back a poor pronunciation. Taking in the estimated quality of detected periodicity into consideration, the hand wrist signalled giving the highest estimated quality factor of the different placements. Adjustments in prototype were made as well since high latency in combination with sudden tempo changes were encountered and a confusion from the participants was expressed. The frame size of N=200 samples was Figure 6: Results of when the participants got to rate how therefore changed to N=100 samples. Several values engaged in the music they were in the first resp. second session. were furthermore experimented with regarding the size Participants who had given a lower rating (2 – little of subset when taking the moving average and the bpm change. Lastly, the functionality regarding how the engaged) would give the motivation that dancing alone bpm should change was modified. Earlier prototype in a room made them feeling little engaged as to incorporated the bpm to be “locked” within the range whenever he/she would be in a club-like situation. of ± 5bpm of the song’s actual bpm if the participant “I think it was not easy to be dancing alone” – P1 would be moving faster/slower than before. This was modified to function without any “lock” within a range “I was all alone. Dancing is kind of a social experience and instead keep increasing/decreasing ± 3bpm for me.” – P8 depending on if the participant is moving faster/slower

Latter mentioned participant (P8) gave this first session a little bubble in which a conversation with myself a rating of 2 (little engaged) while the second session could happen” – P5 got a rating of 5 (completely engaged). For the reason that: 4.2.2 Enjoyment when dancing Figure 7 shows the result when asked to compare their “I can’t say that my dancing improved, but it was enjoyment of dancing between the first and second really engaging when you could control the tempi with session. The x-axis represents the rating from 1 (=much your movements.” – P8 worse) to 5 (=much better). The y-axis represents the Another participant who stated that they gave a number of participants giving a certain rating. different rating the second session (value of 4 – very Following results gave a mean value rating of 3.08. engaged), in comparison to the first session (value of 2 – little engaged) explained the difference as: “Because I was in control of the music. It made it more of a "game" than just dancing to music.” – P6

In relation to the first session, where the motivation for the rating was explained as:

“Hard to "lose yourself" in the music when you are alone in room like this. You feel watched even though the room is empty. It's easier to dance when you're in Figure 7: The rating when comparing the participants’ a room full of people dancing OR at home, where you enjoyment of dancing between the first and second session. feel completely relaxed.” – P6 When asked about how they felt being able to modify Another participant who had given a higher rating the the tempo, a majority described a positive feeling. But first session (value 3 – moderately engaged) but had a along with a positive feeling, some still expressed different engagement the second session (value 2 – having a somewhat split feeling about the interaction. little engaged) explained it as: “It was playful, an enjoyable negotiation.” – P4 “…the changing tempo made it hard for me to enjoy “It was interesting and fun. At the same time there was the music and dance to it. As I moved to the pace of the a feeling of responsibility towards the tempo in music it somehow didn’t catch my movement and comparison to the music that is not tempo modified. began to slow down which made me have to wave my Like if it would be my ‘mistake’ if the dance floor died.” arm fast to make the music speed up again […] on the – P2 other hand, I felt like I got to interact with the music in a new way. The ability to adjust the music as if I was “It felt interesting but exhausting to keep up as my arm dj'ing was cool, as I could play with it. The songs also had to stay in one tempo even though my body might sounded cool when switching the tempo.” – P3 have wanted to working in contradiction to the music sometimes. But it was also interesting to hear when I Other participants who had shifted from feeling got tired and then realising that I had physically moderately engaged to either very or completely changed tempo.” – P11 engaged gave the following reasons: “Great but confusing at times. It made me move in a “The possibility to change the dynamic through my certain way to be sure to not mess up the tempo. Felt a movement was for me more exciting. As well as no need bit restricted.” – P12 to stay repetitive.” – P4 A common feeling of restriction as the last-mentioned “because my moves and actions had an impact on the could be identified among other participants as well. source/reason why I was originally moving. It created

9 “It was hard to use it, as there was a delay of a few highest estimated quality factor of detected periodicity seconds, and as I normally adjust to the beat and have would determine which detected frequency to be used a difficult time to set the pace for it to play, as I then for tempo manipulation, as can also be seen in Figure need to move faster than the music.” – P3 8 and 9. “It slowed down to easily in my opinion. I often felt that the tempo was perfect, but it always slowed down a few seconds later.” – P6 “[…] I felt more engaged in one way, because I could control the pace of the music. Although I felt more restricted because I had to be more repetitive and less instinctual.” – P9 A couple participants described a feeling of uncertainty in if he/she is doing right. “[…] you realise that you sense of beat has gotten worse” – P7

“[…] perhaps a slight misconception from my part Figure 8. Screenshots of recorded sensor data graph, detected when the vocals kicked in as I didn’t feel that I had as frequency and the axis with highest estimated quality of detected periodicity during the session with manipulation from much control over them.” – P10 P9. 4.2.3 Spontaneous reactions During the both sessions of the main study, all participants gave consent to be video-recorded. The objective of recording video during study was to make observations of the spontaneous reactions when being able to control one component of a music playback, parallel with collecting data of their own impression of it expressed in their own words. Recognized among the participants was how a larger part of the participants generated more arm movements during their second session in comparison to their first session. Moving the arm as an indication of exploring possibilities in the tempo manipulation, going from one extreme to another (fast/slow, periodic/non-periodic) was often followed by reacting with a laughter. There were Figure 9. Screenshots of recorded sensor data graph, detected furthermore participants who appeared to shift between frequency and the axis with highest estimated quality of adjusting their dance to what they were hearing and detected periodicity during the session with manipulation from interfere their dance by generating arm gestures to P12. control the tempo. 5. DISCUSSION 4.2.4 Display of sensor data The aim of this study was to evaluate the proof-of- Figures 8 and 9 are two examples of how the signal concept prototype, with the system design proposed in from the sensor data could be displayed when a this thesis work. Real-time measurements were participant could be moving in an intense and repetitive analysed for finding rhythmic movements of a user’s manner versus a participant who could move in a fluid body, which were then mapped to manipulate tempo in and not so regular manner. The axis with the resulting the music playback. The emphasis in the user-study and resulting evaluation was put on the participants’

subjective experience, both with and without the ability the user to create faster movements for a longer time to control the tempo of the music. period in order to make it noticeable. Even though not any statistical significance can be As a result, it becomes more difficult for the user to proven, the results indicate a positive dance experience stay at a desired tempo seeing that the user has to keep for the participants. For a number of them even better, generating movements which increase in frequency for in comparison to when it was a standard EDM music even arriving at a desired tempo. After having arrived playback. at the desired state, the frequency of repetitive movements is then probable to decrease. As was the Preserving the sound quality of the input audio was one concern for P3 who felt that the system didn’t catch the of the main challenges. Granted that the prototype pace of the movements and therefore couldn’t enjoy aimed to strictly change the bpm of the playing input the dance as much. audio – without affecting other sound parameters – the music’s characteristics and the “sound” of it were still 5.1 Method critique affected. A possible explanation for how most of the participants felt more engagement and/or enjoyment Despite an overall positive reaction to the interaction, could be grounded in a feeling that it is music they various factors could have played a part for the study themselves somewhat created. Even though the audio that affected the resulting outcome. Main determinant playback consisted of pre-recorded music, them to consider is the fact that the participants were modifying solely the tempo enabled to contribute to acquaintances, but moreover, a majority were educated sound aesthetic different from the pre-recorded music or are professional dancers. For participants who’ve and another form of participating in the music making. had a long background in dance performance, it is Thus, an enhanced feeling in their engagement and/or assumed that taking on the task given in this study can enjoyment in the interaction. Likewise, it could be happen effortlessly and possibly feel more engaged to originating from the fact that they had to control it more any music that is played for them. Having a larger as and to a greater extent be more attentive to the details well as more diverse population, the distribution of in the music aesthetics. feedback given from the participants could have differed and in turn possibly been applicable for the Modifications that became significant for the user generic public. However, making use of test people interaction were found to be situated in the who are used to work with their body and dance functionality on how the bpm changes. Current system performance in their professional life are also people is designed with the approach of not having too sudden who possess an ability to verbalize what is their know- changes but still within a short time span for the user how which shouldn’t be neglected. to detect immediate responses. A consequence is that it might not be as noticeable for the human ear. While 5.2 Future research speaking in a pure technical context, a change is Improvements and extensions to the proof-of-concept happening. is suggested to be further developed. For example, an Among the participants’ expressed opinions, the most autonomy in the GUI of the prototype could be common quality to be remarked was the delay. In the enhanced. Whereas of now, many steps have to be state that the current prototype is constructed, the performed manually to initiate a session and in the detected frequency is always analysed comparing transition between pieces of music. Additional current subset to previous subset. If the frequency is attempts with modifying the prototype as well as a analysed to be higher, the tempo will increase. If lower, larger number of participants are also encouraged for the tempo will decrease. If the participant is moving better accuracy. “off-beat” or the movements were to be fluid/non- As a continuation to this study, there are a few ideas I repetitive – creating an irregularity – the tempo would think is worth mentioning. Some of the participants decrease. Thus, attempting to make a tempo change for expressed a slight confusion to what sound parameters a short time period might not be heard besides in that in the musical structure that was interacted with, even current subset or coming subset of. It therefore requires though it was limited to tempo alone. Further study on

11 [2] Birgitta Burger, Justin London, Marc R. Thompson and Petri Toivainen. 2018. what other manipulations based on gestures can Synchronization to metrical levels in music depends on low-frequency spectral therefore be explored. Studying how a beat components and tempo. Psychological Research, 82(6), 1195-1211. synchronization would influence the interaction if it [3] Antonio Camurri, Corrado Canepa, Nicola Ferrari, Maurizio Mancini, Radoslaw were to be done more “musically” can be of interest, Niewiadomski, Stefano Piana, Gualtiero Volpe, Jean-Marc Matos, Pablo Palacio and Muriel Romero. 2016. A system to support the learning of movement qualities in e.g. changing tempo and be kept for a certain number dance: A case study on dynamic symmetry. Proceedings of the 2016 ACM of bars (measures). This can provide a solution for the International Joint Conference on Pervasive and Ubiquitous Computing, 973-976. user to feel more engaged in the dance if wanting to [4] Alain de Cheveigné and Hideki Kawahara. 2002. YIN, a Fundamental Frequency Estimator for Speech and Music. The Journal of the Acoustical Society of break off from the interaction and stay in the tempo. America, 111(4), 1917-1930. Investigating other possible interaction mechanisms to [5] Brigid M. Costello. 2018. Rhythm, Play and Interaction Design (Springer Series happen if a clear signal is not received from the motion on Cultural Computing), p. 21, 34. data – and as a consequence deactivate the tempo [6] Sofia Dahl and Anders Friberg. 2007. Expressiveness of a marimba player’s body manipulation – is also suggested. movements. Tmh-Qpsr, 46(1), 075-086.

Investigating how participants would interact if put in [7] Sofia Dahl and Anders Friberg. 2007. Visual Perception of Expressiveness in Musicians' Body Movements. Music Perception: An Interdisciplinary Journal, 24(5), groups can too be worth exploring, and could add the 433-454. dimension of how we socially relate in this type of [8] Paul Dourish. 2001. Where the Action Is. Retrieved from: interaction. https://www.dourish.com/embodied/.

[9] Ludvig Elblaus, Maurizio Goina, Marie-Andreé Robitaille and Roberto Bresin. 6. CONCLUSION 2014. Modes of Sonic Interaction in Circus: Three Proofs of Concept. Proceedings of the Sound and Music Computing Conference, 1698–1706.

The initial objective of this study was to investigate the [10] Steven Gelineck and Niels Böttcher. 2012. 6to6Mappr: An educational tool for subjective experience when a user is let to take control fast and easy mapping of input devices to musical parameters. Proceedings of the 7th Audio Mostly Conference, 117-123. of the decision-making in the music that is played for them. A proof-of-concept prototype was built and [11] Garth Griffin, Youngmoo E. Kim and Douglas Turnbull. 2010. Beat-Sync-Mash- Coder: A web application for real-time creation of beat-synchronous music examined by a total of 12 participants. A user-study mashups. Acoustics Speech and Signal Processing (ICASSP), 2010 IEEE International was conducted consisting of two sessions, one without Conference on, 437-440. the prototype and one with the prototype. [12] Jenny Hansdotter. 2003. Audio Feedback in music – a study of experience of audio feedback in music for rehabilitation treatment to fear avoidance. Master’s thesis. Royal Institute of Technology (KTH), Stockholm, Sweden. The proposed design is suggested to provide a dance experience that can compare positively to a standard [13] Jason A. Hockman, Marcelo Wanderley and Ichiro Fujinaga. 2009. Real-Time Phase Vocoder Manipulation by Runner’s Pace. In Proceedings of the 2009 playback of EDM music. Results imply giving an Conference on New Instruments for Musical Expression, 90-93. overall positive dance experience worth exploring [14] Wolfgang Hürst, Ronald Poppe and Jerry Van Angeren. 2015. Drawing outside further. For a number of the participants, the prototype the lines: Tracking-based gesture interaction in mobile augmented entertainment. Intelligent Technologies for Interactive Entertainment (INTETAIN), indicated contributing to more engagement and 2015 7th International Conference on, 79-87. enjoyment than to a standard playback of EDM [15] Alexander Refsum Jensenius, Marelo M. Wanderley, Rolfe Inge Godøy and Marc involving not interacting with the prototype. Leman. 2010. Musical gestures: Concepts and methods in research. In R. I. Godøy & M. Leman (Eds.), Musical gestures: Sound, movement, and meaning. New York, NY, USA: Routledge, 12-35. 7. ACKNOWLEDGMENTS [16] Youngkee Jung and Byungrae Cha. 2010. Gesture Recognition Based on Motion I would like to thank all my bold participants who were Inertial Sensors for Ubiquitous Interactive Game Contents. IETE Technical Review, 27(2), 158-166. willing to expose themselves in front of a video camera [17] Marc Leman and Luiz Naveda. 2010. Basic Gestures as Spatiotemporal Reference to me, I know it can’t have been an easy thing to do for Frames for Repetitive Dance/Music Patterns in Samba and Charleston. Music all of you. I would also like to thank my supervisor Perception: An Interdisciplinary Journal,28(1), 71-91.

André Holzapfel for the great input and guidance [18] Bart Moens, Chris Muller, Leon Van Noorden, Marek Franěk, Bert Celie, Jan throughout the project. Boone, Jan Bourgois and Marc Leman. (2014). Encouraging spontaneous synchronisation with D-Jogger, an adaptive music player that aligns movement and music. PLOS ONE, 9(12), E114234. 8. REFERENCES [19] Pieter-Jan Maes, Denis Amelynck, Micheline Lesaffre, Marc Leman and D. K. [1] Frederic Bevilacqua, Fabrice Guédy, Norbert Schnell, Emmanuel Fléty, and Arvind. 2012. The “Conducting Master”: An interactive, real-time gesture monitoring Nicolas Leroy. 2007. Wireless sensor interface and gesture-follower for music system based on spatiotemporal motion templates. International Journal of Human- pedagogy. In Proceedings of the international conference on new interfaces for Computer Interaction, 29(7), 471-487. musical expression, New York City, NY, United States, 124–129. [20] Luiz Naveda, Isabel Martínez, Javier Damensón, Alejandro Ghiena and Romina Herrera. 2015. Methods for the Analysis of Rhythmic and Metrical Responses to

Music in Free Movement Trajectories. In 11th International Symposium on Computer Music Multidisciplinary Research. CMMR, Plymouth, UK.

[21] Radoslaw Niewiadomski, Maurizio Mancini, Stefano Piana, Paolo Alborno, Gualtiero Volpe and Antonio Camurri. 2017. Low-intrusive recognition of expressive movement qualities. Proceedings of the 19th ACM International Conference on Multimodal Interaction, 2017, 230-237.

[22] João Lobato Oliveira, Fabien Gouyon, Luis Gustavo Martins and Luis Paulo Reis 2010. IBT: Real-time tempo and beat tracking system. In Proceedings of International Society for Music Retrieval (ISMIR).

[23] João Lobato Oliveira, Matthew E. P. Davies, Fabien Gouyon and Luís Paulo Reis. 2012. Beat Tracking for Multiple Applications: A Multi-Agent System Architecture With State Recovery. Audio, Speech, and Language Processing, IEEE Transactions on, 20(10), 2696-2706.

[24] João Lobato Oliveira, Gökhan Ince, Keisuke Nakamura and Kazuhiro Nakadai. 2012. Online audio beat tracking for a dancing robot in the presence of ego-motion noise in a real environment. Robotics and Automation (ICRA), 2012 IEEE International Conference on, 403-408.

[25] Andrew Robertson and Plumbley. 2007. B-Keeper: A Beat Tracker for Live Performance. In Proceedings of New Interfaces for Musical Expression, 234-237.

[26] Andrew Robertson and Mark D. Plumbley. 2013. Synchronizing Sequencing Software to a Live Drummer. Computer Music Journal,37(2), 46-60.

[27] Catarina B. Santiago, João L. Oliveira, Luís P. Reis, Armando Sousa and Fabien Gouyon. 2012. Overcoming Motor-Rate Limitations in Online Synchronized Robot Dancing. International Journal of Computational Intelligence Systems, 5(4), pp 700- 713.

[28] Ragnhild Torvanger, Solberg and Alexander Refsum Jensenius. 2017. Pleasurable and Intersubjectively Embodied Experiences of Electronic Dance Music. Empirical Musicology Review. 2017, 11(3-4), 301-318.

[29] Ragnhild Torvanger Solberg and Alexander Refsum Jensenius. 2016. Optical or Inertial? Evaluation of Two Motion Capture Systems for Studies of Dancing to Electronic Dance Music. Logos Verlag Berlin.

[30] Atau Tanaka. 2011. Music one participates in. Proceedings of the 8th ACM Conference on Creativity and Cognition, 105-106.

[31] Petri Toivainen, Geoff Luck and Marc R. Thompson. 2010. Music Perception: An Interdisciplinary Journal, Vol. 28 No. 1, September 2010, 59-70. DOI: 10.1525/mp.2010.28.1.59

[32] Vibra. 2018. Connecting multiple sensors part 1. Retrieved from https://www.vibra.no/home-1/connecting-multiple-sensors-part-1-ngimus-over-wi-fi.

[33] Federico Visi and Luke Dahl. 2018. Real-Time Motion Capture Analysis and Music Interaction with the Modosc Descriptor Library. NIME’18, June 3-6, 2018, Blacksburg, Virginia, USA.

[34] Federico Visi, Eduardo Reck Miranda, Esther Coorevits and Rodrigo Schramm. 2017. Musical instruments, body movement, space, and motion data: Music as an emergent multimodal choreography. Human Technology, 13(1), 58-81.

[35] Qi Wang, Panos Markopoulos, Bin Yu, Wei Chen and Annick Timmermans. (2017). Interactive wearable systems for upper body rehabilitation: A systematic review. Journal of NeuroEngineering and Rehabilitation, 14(1), 20.

[36] Wikipedia. 2019. WikipediA: The Free Encyclopedia. Retrieved from: https://en.wikipedia.org/wiki/Electronic_dance_music

[37] Yang Zhao, David Gerhard, Markus Brahms and John Barden. 2016. Stance phase detection for walking and running using an IMU periodicity-based approach. Advances in Intelligent Systems and Computing, 392, 225-232. DOI:10.1145/1234567890

13 A. Resulting Max/MSP-patches8

Figure A.1. Screenshot of Max/MSP-patch in presentation mode

Figure A.2. A screenshot of step 4 in the chain of operations illustrated in Figure 2. The attributes in the figure were the attributes used for the yin-object. These could be modified manually or to be static when using the patch.

8 Full patch available at https://github.com/lilianjap/periodicity_detection 14

Figure A.3. The detected values in Figure A.4 will be sent to this patch, where the values are analysed and converted to corresponding bpm-value.

15 B. List of used songs

1. Abba 002 - Axel Boman 2. ATM jam (Kaytranada Edition) - Azealia Banks 3. Baby I'm yours - Breakbot ft. Irfane 4. Be Your Girl (Kaytranada Edit) - Teedra Moses 5. Boy - HNNY 6. Chance the Rapper, Knox Fortune - All Night (Kaytranada Extended Joint) - Kaytranada 7. Cole's Just Crazy - Cole Medina 8. Deep Inside (the Dub) - Hardrive 9. Everything is Everything (Krystal Klear Remix)- Sky ferreira 10. Hey, What's Wrong With You - Armand Van Helden 11. I Can't Get No Sleep (Ken Lou 12'') - Masters At Work ft. India 12. If (Kaytranada Edit) - Janet Jackson 13. Is It All Over My Face - Loose Joints 14. J.A.W.S. - Lxury 15. Juno - Genius of Time 16. Let No Man Put Asunder (Ron Hardy Mix) - First Choice 17. Love thang (Genius of Time edit) - First Choice 18. Myrtle Music (Original Edit) - Earth boys 19. Parental Advisory House - House Is Ok 004 Janis 20. Pusherman (BONNIE EDIT) - Curtis Mayfield 21. Raingurl - Yaeji 22. Running (Disclosure Remix) - Jessie Ware 23. Saturday Love (Vandalized edit) - Jarreau Vandal 24. Secret Fantasy - James Sims 25. Shake it to the Ground - Rye Rye 26. Solange - Cranes In The Sky (Kaytranada DJ Edit) 2 - Kaytranada 27. Tears - HNNY 28. The Weeknd x Drake - Crew Love (Star Slinger's Jersey Refix) - Star Slinger 29. This Time Baby - Jackie Moore 30. Track 7 - Jai Paul 31. U Don't Know Me (ft. Duane Harden) - Armand Van Helden 32. Ultimatum (ft. Fatoumata Diawara) - Disclosure 33. Why don't we fall in love (Kaytranada Edit) – Amerie

Pop C. Questionnaire + Responses: i like to listen to Rock, House, Metal, Rap. My year in Spotify was a bastard child of all kinds of genres. C.1 Demographic information Electronic - Vapor Wave (sort of) EDM, electro, pop Gender M 8 (56,7%) C.2 First Session F 4 (33,3%) How engaged in the music did you feel you were in the Prefer not to say 0 (0%) session? Scale 1-5 (Not at all engaged-Completely engaged) Age: See Figure 6.

What made you give that rating? i think it was not easy to be dancing alone I lost track a couple of times. Started thinking about the instructions I got before the session started. Maybe mostly in the transitions between the songs. But as able to pick i up and back to it. i did not feel entirely comfortable in what i am wearing at the moment. however i likedthe music and felt like moving Any dance experience? to it. Yes 9 (75%) Shifting from 2-5 during the 10 min session, 3 being No 3 (25%) somehow average other tasks that i had to keep in mind & increadsed volume If yes, what is your previous experience? could maybe push rating up. i am a professional dancer Hard to "lose yourself" in the music when you are alone in Dancing school, work as a dancer room like this. You feel watched even though the room is streetdancer since 11 years of age, teaching, competing and empty. It's easier to dance when you're in a room full of judging event. people dancing OR at home, where you feel completely Professional dancer relaxed. dancer couldv'e been much more extra, but it was a long time ago Danced in my earlier years, some street and one year of since i danced and wasn't all too relaxed for being that ballet I was all alone. Dancing is kind of a social experience for Been working with dance for the past 10 years me. Working professionally for 9 years as a freelancing dancer I was letting the different rythems, moods and the pace of in the field of contemporary dance, ballet, baroque dance the music affect my dancing. and performance. I just closed my eyes after a while and imagined myself Profession being in a club on the dancefloor I really enjoyed myself with the music but I didn't have a Any genre in particular you like to listen to? euphoric surrendering to it as sort of total engagement. eletro music and pop music Usually not into too much house, but it was groovy house EDM, EBM, Soul, jazz, metal, pop, world music, andean so it made me engage. music, r'nb, hip-hop, punk soul, funk, cumbia, hiphop and some electronic music C.3 Second session Various How engaged in the music did you feel you were in the session? Scale 1-5 (Not at all engaged-Completely fan of annything that is sincere engaged) All-eater. More live instruments than electronic music. See Figure 6. Hip hop, pop, classical pop, techno, rock

17 Was there any difference in the rating, in comparison to no difference to be honest, yes I had more control over the previous session? If yes, could you describe to why you music, but I still felt the same level of engagement, just gave a different rating? closing my eyes and going ham! yes it was a big difference because i could not let go and I rated the same. But for different reasons. First time I was just enjoying dancing and anticipate the music , in this case more engaged with the differences and nuances in the i was the leader of the rythme and for me it was harder to composition. While this second time I had to engage and follow the rythme wih is very paradoxal , i was a bit relate more directly physically to the music to keep it in frustrated and less free in my dance bu i found it very places I wanted it to. Also I had to react more as the music interested and challenging was changeing tempo. But I felt I was not listening as much Same rating. But very exciting to try to figure out how the to the music and mostly to the beat. music responded to the movement at the same time as The power of managing the tempo frustrating to at moments to have the music allow. I guess that as a dancer I''m used to be able to go "against" the How did the tempo modification felt to use? music. it was good and enjoyable yes, the changing tempo made it hard for me to enjoy the It was interesting and fun. At the same time there was a music and to dance to it. as i moved to the pace of the feeling of responsability towards the tempo in comparison music it somehow didnt catch my movement and began to to to music that is not tempo modified. Like if it would be slow down , which made me have to wave my arm fast to my "misstake" if the dnce floor died. make the music speed up angain, which took my focus it was hard to use it, as there was a delay of a few seconds, harmony and presence away. also, the sound of the music and as i normally adjust to the beat and have a difficult time slowing down felt symbolic to me, as if me/obervers were to set the pace for it to play, as i then need to move faster falling asleep becasue of my repetitive and uninspired than the music. dancing. also, the delay of a few seconds made it more hard It was playful, an enjoyable negotiaion to manipulte the music the way i wanted, so i mostly felst Fun, not responsive enough like i was fighting fo it to even play. on the other hand, i It slowed down to easily in my opinion. I often felt that the felt like i got to interact with the music in a new way. the tempo was perfect, but it always slowed down a few abibilty to adjust the music as if i was dj'ing was cool, as i seconds later. could play wiht it. the songs also sounded cool when scary, you realise that you sense of beat has gotten worse swithcing the tempo. It made me move my right arm much more. I imagine if The possibility to change the dynamic through my you hade more sensor this would apply to the whole boy. movement was for me more exciting. As well as no need to Overall it felt good, kind of responsive with a bit of delay. stay repetative Great! because my moves and actions had an impact on the source very easy, perhaps a slight misconception fomr my part / reason why i was originally moving. it created a little when the vocals kicked in as I didn't feel that I had as much bubble in which a conversation with myself could happen. control over them Because I was in control of the music. It made it more of a It felt interesting but exchausting to keep up as my arm had "game" than just dancing to music. to stay in one tempo even though my body might have I feel that this was much more difficult, but i tried hard but wanted to working in contradiction to the music sometimes. not the best i could (as prev.) But it was also interesting to hear when I got tired and then I can't say that my dancing improved, but it was really realising that I had physically changed thempo. engaging when you could control the temp with your Great but confusing at times. It made me move in a certain movements. way to be sure to not mess up the tempo. Felt a bit No, I felt more engaged in one way, because I could restricted. controll the pace of the music. Altough I felt more restricted because I had to move more repetative and less How did your enjoyment of the dance change between instinctual. the first and second trial? Scale from 1-5 (Much worse- Much better) See Figure 7.

Are there any type of sound modifications you think none besides that I really enjoyed this would have felt better? Would be interesting to try to dance with somebody else No that wears the tracking device to see how the social and no because it already a lot to embrace and understand physical relation between our bodies would have to adapt Don't know of any to that persons movemens. maybe different areas in the room, where you move closer No. and it plays faster or louder (well, thats the normal effect if you move closer to the speakers hehe) another thing might be if the pace-setter was more trimmed to follow your body and adjusted more quickl to you. that would have made it feel less frustrating to have to "pump up" the pace, as it would be more of a flow. More sensetinve and responsive equipement. Maybe with filters. more sensors at different boddy parts with different functions. Changing the placement of the sensor controlling the tempo. I don't move my hands to the beat of the music when I dance. The feet would probably have worked better for me, I think. Like I said, the tempo slowed down to easily. the first session was easier and much more relaxed, since i didnt have to decide the tempo..... It would be cool if you could change the pitch depending on height from the ground. no not in this case, when you dance I ought to believe that you want the control over the tempo not the volume or the panning I really enjoyed the feeling of being able to change the tempo as I like it during he song/music. But I think I would have prefered to be able to press a button for the tracking to start adapting tempo to my movement, and later that it keeps the tempo that I had when I stop pressing the button. To release the pressure of keeping working with the arm to keep the tempo in the right mood. Amount of instruments would be interesting to explore. Like "depths" in the sound.

Any other comments? No i wish it was no delay between my mouvement and the tempo no tack!! det var kul att testa :) i want to do more ! It was fun! 'TWAS FUN!!! good luck x I just think that if the delay could be fixed, it would be much more engaging.

19 TRITA -EECS-EX-2019:88