CN101120343A

CN101120343A - Electronic device and method for selecting content items

Info

Publication number: CN101120343A
Application number: CNA2006800048517A
Authority: CN
Inventors: S·C·波维斯; F·维格诺利; G·L·维纳尔达
Original assignee: Koninklijke Philips Electronics NV
Current assignee: Koninklijke Philips NV
Priority date: 2005-02-14
Filing date: 2006-01-30
Publication date: 2008-02-06
Anticipated expiration: 2026-01-30
Also published as: JP5149017B2; JP2008530678A; EP1851661A1; WO2006085237A1; CN101120343B; KR101333024B1; KR20070112189A; US20080188354A1

Abstract

The method of selecting content items of the invention comprises selecting (6) a subset of content items from a collection of content items in dependency of feedback gathered during a previous physical activity of a person (3). The subset is smaller than the collection of content items and the feedback comprises information regarding the previous physical activity. The electronic device of the invention comprises electronic circuitry operative to perform the method of the invention. The software of the invention enables a programmable device to perform the method of the invention.

Description

Electronic device and method for selecting content items

Technical Field

The invention relates to an electronic device for selecting content items, for example a music player or a device for transferring songs to a music storage or a music player.

The invention further relates to a method for selecting a content item.

The invention also relates to software for causing a programmable device to perform a method of selecting a content item.

Background

An electronic device for reproducing human perceptual signals is known from WO 2004/072767. WO2004/072767 describes an audio reproduction apparatus which modifies a tempo (tempo) of an audio signal such that a deviation of the modified tempo of the audio signal from a selected tempo is within a predefined acceptable range. The selected tempo may be based on a human heart rate, pace (pace), and/or a user-specified tempo. A piece of audio may be selected from the audio source either manually by the user or automatically by the audio reproduction device itself (based on the selected tempo). A drawback of the known audio reproduction apparatus is that the selected audio piece is often not optimal for the physical activity the user is performing, especially if the user has to transfer the audio piece to the music (player) memory in advance of the start of the physical activity, when the memory is relatively small.

Disclosure of Invention

It is a first object of the invention to provide an electronic device of the kind described in the opening paragraph, which provides a better selection of content items for rendering during physical activity.

It is a second object of the invention to provide a method of the kind described in the opening paragraph which provides a better selection of content items for rendering during physical activity.

The first object is according to the invention realized in that the electronic device comprises an electronic circuit for selecting a subset of content items from a set of content items, the subset being smaller than the set of content items and the feedback comprising information about a previous physical activity of the person based on feedback collected during the previous physical activity. By using feedback in the content selection step, which contains information about previous physical activities of the person, a content item can be selected, the rendering of which results in the most beneficial performance of the person. The characteristics of the content item whose rendering results in the most beneficial performance of the person are part of the feedback or can be derived from the feedback. Even if it is possible for a user to manually select an appropriate content item, this requires a great deal of effort. The invention is particularly advantageous when the user has to transfer a subset of content items to a music (player) storage before physical activity.

For example, the content item may contain audio, video, light and/or haptic feedback. The electronic device may be, for example, a personal computer or a portable rendering device for transferring the subset of content items to a memory, the portable rendering device itself (e.g. a mobile phone) or a stationary rendering device (possibly integrated into some kind of fitness equipment). The collection may be a private collection (private collection) or a public collection (public collection) (e.g., a collection of content providers). For example, the electronic device may be used at a health club, at home, or outdoors. The subset of content items may be stored locally or remotely (e.g., at a content or customer service provider).

The feedback may comprise the physical state of the person in a previous physical activity. This is particularly useful when the person has to reach a target pace, but the content item is synchronized with the physical state of the person (e.g. heart rate). The physical state may comprise, for example, a heart rate of the person. The physical state may be, for example, an average physical state of the person over a plurality of exercise sessions, or an average physical state over each exercise session, or an average physical state of the person over each of a plurality of time periods.

The feedback may include the pace of the person during previous physical activity. This is particularly useful when the person has to reach a target physical state (e.g. heart rate), but the content item is synchronized with the person's pace. The pace may include, for example, revolutions per minute while riding a bicycle or steps per minute while jogging. The pace may be, for example, the person's average pace over a plurality of workouts, or the person's average pace over each workout, or the person's average pace over each of a plurality of sessions.

The feedback may comprise a pace associated with a physical state of the person during a previous physical activity. This is particularly useful when the tempo of the audio output signal depends on both the pace and the physical state of the person. The health of the person and thus the physical state of the person are dynamic when a certain pace is reached: it is usually dependent on the recent amount of physical activity of the person.

The feedback may comprise information marking which ones of the subsets of content items have been skipped during a certain type of physical activity. Although a certain content item may seem to be suitable for a certain type of physical exercise based on its tempo or the user's usual preferences, the result may prove that it is not suitable for said type of physical exercise and may in fact adversely affect the performance of the person. The certain content item will not be selected for the certain type of exercise and may not even be transferred to the music (player) memory.

The feedback may comprise information identifying an impact of one or more of the subset of content items on a physical state of a person. Some content items may have an unexpected beneficial effect on the physical state of a person (i.e. the performance of the person). These content items will be selected preferentially.

The feedback may include the length of time of a certain type of physical exercise. A certain type of exercise may need to be prolonged because the goal of the exercise has not yet been reached, or the certain type of exercise may be automatically prolonged because the person likes this type of exercise. When the user has to transfer a subset of content items to a music (player) memory before the start of a physical activity, and the memory is relatively small, the exact length of time of the workout is important information for selecting an appropriate amount of content items for each workout,

the second object is according to the invention realized in that the method comprises the step of selecting a subset of content items from a set of content items on the basis of feedback gathered by the person during a previous physical activity, the subset being smaller than said set of content items and the feedback comprising information about the previous physical activity.

Drawings

These and other aspects of the electronic device and method of the invention will be further elucidated and described with reference to the drawings, in which:

FIG. 1 shows an example of a fitness program (Workout scheme) for use in an embodiment of the method of the present invention;

FIG. 2 is a flow chart of a method embodiment of the present invention; and

fig. 3 is a block diagram of an electronic device of the present invention.

Corresponding elements in the drawings are identified by the same reference numerals.

Detailed Description

The use of music can be of interest for physical exercise, especially for individual, repetitive and endurance sports. Listening to music before or during exercise is believed to increase motor output by mental separation, motivation, and by improving emotional state, and is believed to provide improved control and learning motor controlled skills. Thus, music can be an excellent guide to keep up with physical exercises (physical exercise), especially when such physical exercises are monotonous, repetitive and boring. It has become a daily occurrence on the street that runners wear headphones to listen to music.

Many recreational exercisers and endurance athletes/sports enthusiasts work with personal training programs directed at their fitness and athletic experience. These training programs basically define the sequence of exercises that need to be performed, as well as the duration and intensity of each individual exercise. They fit into a complete plan and are modified over time either to strive for optimal performance, weight control, or recovery from injury.

The method of the invention is preferably used in a personalized music system for physical exercise that enables integration of a physical training program, music selection and music playback (playback). It will make the physical exercise more enjoyable and will help the user keep up with the intended goals set in the training program. Briefly, prior to exercise (work out), the system supports a user to pre-select preferred music appropriate for physical exertion during a workout routine. During fitness, a song playback is selected and the nature of the music playback, in particular the tempo of the music, is modified to reflect or guide the current user performance in terms of heart rate and exercise frequency. After fitness, performance data and playback data are collected for visual inspection, for training program modification, and for refining future music through learning.

Even when these physical exercises induce feelings of monotony, boredom, pain, fatigue, and exhaustion without providing immediate rewards, many individual endurance exercise exercises require motivation to causeThe person insists on going down. In the field of sports and exercise psychology, it is assumed that exercisers use a variety of joint/internal and separate/external strategies to cope with the actual demands of physical strength and required endurance due to sports exertion (sport). With less than moderate physical effort loads, exercisers are able to automatically divert attention (i.e., "detach") to internal sensations from the muscle, joint, cardiovascular and pulmonary systems. The separation strategies used are image creation (imagediscovery), positive mood development (e.g. by recalling what is nice in life), consideration of work, career or social relationships, creation of self-efficiency (self-confidence), self-esteem and confidence, or problem solving such as mental arithmetic. Obviously, at about 70% of Maximum Heart Rate Reserve (MHRR), exercisers are forced to start diverting their attention from external cues back to the internal sensation of physical discomfort. At the degree of physical load at which extreme fatigue or exhaustion is felt (e.g., at 90% ₂ MAX), exercisers are no longer able to "decouple" themselves from the imminent painful sensations. Instead, exercisers must "fight" directly with the pain in order to adhere to their exercises, which requires a high level of motivation and exercise tolerance.

In addition to making physical exertion more enjoyable, music may help exercisers to escape from the feelings of exertion or even become a motivator for continued effort. Studies have shown that fast music (up-tempo) can improve performance, but only on physical tasks with low and moderate levels of exercise(i.e., with less than maximum effort) is effective. Doubling the music pace at significant 70% mhrr critical levels enables exercisers to postpone their fatigue time; it appears that changes in music tempo help them extend the period of using the "split" coping strategy. If the exerciser is explicitly instructed to synchronize with the music being played, the performance will be improved. In particular, when pedaling speed andwith music synchronized, endurance is improved in a stationary, constant-pace bicycle mission. Background music that is not coordinated with pedaling does not show the same effect as the effect without music. VO at high intensity exercise (e.g., 90% ₂ MAX), the feeling of exertion becomes overwhelming and indifferent, then reaching the level of extreme fatigue and exhaustion. As already mentioned, at that level of exertion, exercisers can no longer actively divert their attention to external stimuli. Thus, while music still brings greater enjoyment to the task than without it, music does not help them improve to sustain their efforts.

The right combination of solo endurance sports and music playback has produced consumer products. The commercial collaboration of the electronic equipment company Philips and the sportswear manufacturer Nike created a combination device called MP3Run (PSA 260) that was used as a portable music player and performance/training display and recording system. The player is a 256MB flash memory device for playing music encoded in approximately 4 to 8 hours MP3 or WMA format, and comprises an FM radio. The music may be updated by downloading. For performance recording, the device is provided with a pedometer that wirelessly transmits (using Bluetooth) speed and distance data to the device for display and on-demand audio feedback. The performance data may be uploaded to a personal computer for review. However, music playback or update is not combined with the presentation.

The training program specifies the sequence of physical exercises scheduled in a single training session. The definition of exercise may include (see fig. 1):

exercise type, which refers to exercise in sports like running, cycling, rowing, aerobic exercise, and fitness;

an exercise duration specifying a time period for performing the exercise, which may be at a prescribed intensity level;

an exercise intensity specifying at what exertion level the exercise should be performed. It may be expressed as a percentage of the Maximum Heart Rate Reserve (MHRR) which may specify a desired heart rate range. For example, aerobic heart rate range for improved endurance and optimal cardiovascular training is 70-80% MHRR;

exercise stringency (exercise severity) which tells how strict or loose the exercise should be in terms of duration and intensity.

For a particular exercise type, the parameters are irrelevant or unpredictable. For example, heart rate during warm-up or stretching exercises is very little related and is assumed to be at rest. It is not known in advance how long it takes to return from an exertional state to a state where the heart rate is calm.

Note that the training program may also include a single workout, such as running an hour, for those who prefer "running easily through a park" rather than having to concentrate on the training program. Typically, the training program is part of a complete and personalized training program for achieving well-defined long-term goals, such as trying to achieve optimal performance in a physical event, extending endurance, strengthening strength (strength and power), losing weight, getting healthier, maintaining health, or rehabilitating the body. These programs, together with these training programs, are created by sports physiologists, coaches, trainers, sports and health counseling companies or by the individual himself, possibly with the aid of modern computer technology.

In the embodiment of the invention shown in fig. 2, the user is helped keep up with the desired goals set in the training program with the favorite music in 4 use phases. The 4 phases are an installation phase (not shown), a preparation phase 1, an exercise phase 2, and a feedback phase 4. The installation phase, preparation phase 1, and feedback phase 4 are offline. In other words, tasks related to these stages are performed using a personal computer at home before and after physical exertion. Conversely, when the user is performing physical exercise, exercise stage 2 is online.

Stage of installation

The first use of the system requires an installation phase (not shown) to personalize the system. In one session, the system requests relevant personal data (e.g., name, gender, age), exercise and physiological data (e.g., weight, fat content, heart rate at rest, maximum or peak heart rate, stride frequency, stride length), and level of athletic experience (i.e., beginner, intermediate, experienced) from the user. In addition, the definition of the exercise and the likes and dislikes of music may be changed or added. The system can be used by multiple users, but all of them have their personal profiles. Users can always return to this installation phase to reset their data.

Preparation phase 1

Preparation stage 1 is associated with off-line training program creation and music preselection. The prepared training plan may provide for use, modification, or may be created from scratch. Next, preferred music is selected from a larger collection of music that fits the training program. The pre-selection of music is performed before the actual training begins, thus ensuring a comfortable-free (hassle-free) sport and musical experience. In addition, it overcomes the storage limitations on current flash-based portable players. Different training exercises have quite different characteristics in terms of movement, pace, intensity and duration. For example, warming-up is different from interval training exercises because it does not consider exercise intensity or user performance. In contrast, warming up requires the full attention of the exerciser. These different exercise features require different favorite music selections. Both music and training programs are downloaded into portable players that can be carried. The following requirements are designed in the music preselection strategy of the system:

personal music preferences related to exercise are considered in the training program, relating to likes and dislikes of songs, music genre, artist, etc.

The amount of music selected is proportional to the duration of the individual exercises as a whole and to the duration of the exercise program.

Music is selected according to the intensity of the exercise. In particular, the exercise intensity is converted into a music tempo property.

A user performance indicator is employed to learn what songs need to be included or excluded in future preselections.

The preparation phase 1 comprises a step 6: as shown in fig. 2, a subset of content items is selected from the set of content items, the subset being smaller than the set of content items, based on feedback collected by the person 3 in a previous physical activity, and the feedback comprising information about the previous physical activity. In the embodiment shown in fig. 2, the feedback is processed in the feedback stage 4. In an alternative embodiment, step 6 may be performed as part of an exercise session performed at the server. A (e.g. known) content rendering device may be connected to that server during the physical activity. In this alternative embodiment, the offline phase is not necessary.

To shape the music preselection process, we define the song as a finite vector s = (v) with K attribute values ₁ ，...v _K ). These attributes reflect tags of catalog data or intrinsic characteristics of music such as unique identifiers, title information, artist name, genre, duration, music tempo, and beat notation. The values for these tags can be easily provided by an online service from a third party or a music signal processing algorithm. All possible values for each kth attribute are centered by D _k The attribute field of the representation.

A finite set of M songs M =(s) ₁ ，...，s _m ) Representing a music collection of a user. Preselection of music is achieved by generating a play set defined as a (unordered) set of n songs p = (p) ₁ ，...，p _K ) Wherein p is for i =1 _i E.g. M. We use p _ik Representing the kth attribute value of the ith song in the play set. For annotation convenience, we consider to beKnowing the size of the play set n. In practice, the size of the play set is made variable。

For each individual exercise, a set of constraints C is defined by the user or system that can generate the play set. In addition, constraints are defined that will apply throughout the training program, for example, to ensure that different songs are selected for these exercises, and that all music fits into a portable player. All play sets are combined into one pre-selected music set to be used during the exercise session. The constraint is a restriction in the declarative terms that tells what songs should be in the play set and what songs should not be in the play set. Constraints may reflect a user's music preferences or restrictions on music attributes suitable for physical exercises in an exercise. If the play set satisfies the limits defined by the constraints, then the play set is deemed to satisfy one of the constraints. Otherwise, it is considered to violate the constraint.

If the user dislikes music with a particular attribute (e.g., artist or genre) while doing a particular exercise, she can inform the system of this fact through the user interface to prevent selection of such music. The system then instantiates an exclusion constraint with a specific instantiation. Formally, the exclusion constraint is given by a four-dimensional number (p, n, k, V), where:

p is a play set, and p is a play set,

n is the size of p and is,

k is an attribute number, K is more than or equal to 1 and less than or equal to K,

v is a set of attribute values, V583976D _k ，

Indicates that for all 1 ≦ i ≦ n, it must hold p _ik ∉V。

On the other hand, if the user likes music with a particular attribute, the system instantiates a counting constraint that limits and enforces the number of occurrences of the particular attribute value in the play set. Formally, the count constraint is given by a six-dimensional number (p, n, k, V, a, b), where:

p is a play set, and p is a play set,

n is the size of p and is,

v is a set of values of the attributes, _k VD _k ，

a and b are the lower and upper limits respectively,

meaning that it must maintain a ≦ i ≦ n | p {1 ≦ i ≦ n ≦ p ≦ _ik ∈V}|≤b。

If the workout has a limited duration, then a group of songs with a similar total duration may need to be selected for which a duration (duration) constraint may be used. Formally, the duration constraint is given by the five dimensions (p, n, k, a, b), where:

p is the play set and is the number of play sets,

n is the size of p and is,

a and b are respectively a lower limit and an upper limit,

indicating that it must hold

An exercise is generated by limiting its intensity and severity to a play set over the range of music tempos that need to be included. If we assume that the frequency of motion will be synchronized with the music (or vice versa), only music with a specific music tempo range is needed, for which a range constraint may be provided. For each particular user, a precise limit on the music tempo relating to the exercise needs to be learned. Formally, the range constraint is given by the five dimensions (p, n, k, v, w), where:

p is a play set, and p is a play set,

n is the size of p and is,

v and w are the lower and upper limits, respectively, where v, w ∈ D _k

For allThe portion 1. Ltoreq. I.ltoreq.n, which must hold v.ltoreq.p _ik W is less than or equal to w. Likewise, songs may be generated for a play set as if the kth attribute value were derived from a statistical distribution (e.g., a normal distribution).

Note that the constraints that define the order of songs are not modeled, as play sets are considered unordered sets.

1) The local search is realized: satisfying a set of constraints is an NP-hard combinatorial problem. It is unlikely that there will be a polynomial algorithm that computes playsets that satisfy any given set of constraints. If constraints conflict, there are even cases where there is no feasible solution. Thus, rather than seeking an exact solution, an approximate solution is computed by using a local search. In this local search method, the constraints on each play set are converted into normalized, intelligent splice linear penalty functions. Each penalty function is defined as zero if the constraint is satisfied, is defined as greater than zero if the constraint is not satisfied, and increases as the number of violations of the constraint increases. For example, a penalty function for a duration constraint is defined as:

wherein

And δ = max (a-n.minD) _k ，n·max D _k -b). If the sum of all the k-th attribute values (e.g., total duration) is in the range a, b]Within, then the penalty is zero. Otherwise, the penalty is the normalized difference between the sum and the nearest lower or upper limit (normalizeddiference).

The total penalty is defined as a weighted convex combination (weighted conjvex combination) of all penalty functions involved. The weight indicates how severe the unsatisfied condition is if the constraint has not been satisfied. Some constraints are very strict, while others represent that they may be partially violated. Selecting too few songs of a favorite artist may be more restrictive than selecting more music than may be suitable for an exercise.

In order to find a play set that minimizes the total penalty, the local search considers either the full solution or the play set. The solution is stepped by applying a random, small change in each iteration. Such changes that may be allowed are: (ii) add a song, (ii) delete a song, and (iii) replace a song with a new song. Since no constraints on the sequence order are mentioned here, there is no great relationship to changing the song order. If the newly generated music selection is better than the original music selection, the new music selection is accepted and the next cycle of the local search is entered. If the new solution is worse, the solution may be accepted with a probability that decreases with the amount of degradation (determination), but also decreases during the course of the algorithm. In other words, the search process "cools down," which is referred to as simulated annealing, to avoid trapping in local minima. A predefined cooling schedule (schedule) needed for simulated annealing for temperature control; a standard linear cooling schedule is employed.

Small modifications to the above approach do not convert all constraints into penalty functions, but some easy constraints are handled by preprocessing, i.e. using constraint propagation principles. For example, if a user wants all songs to be rock songs, one can easily do so by using only rock songs from the music collection without considering all the remaining songs.

Exercise stage 2

Referring to fig. 2, the exercise stage 2 is an online stage in which preselected music is played back as the user exercises. The exercise instructions in the training program will play back what music. The system, which is currently a portable player, monitors and records user performance data such as heart rate using a wireless connection to a heart rate sensor, and exercise frequency (i.e., the runner's step frequency) and running speed using a wireless connection to a pedometer. Depending on the desired mode of actuation, the music playback is adapted to the current user performance or current training goal. In addition to standard linear playback, the following advanced music playback modes have been implemented:

constant tempo (constant tempo): the music is played back at a constant tempo, thereby encouraging the exerciser to synchronize his pace with the music used to enhance endurance. The goal is to help the exerciser distribute his strength evenly over a distance and reduce recovery costs due to brief high intensity exercises.

Pace matching (pace matching): the music tempo is continuously adapted to the exerciser's current stride frequency, thereby enabling running at different stride frequencies without being preoccupied with a continuous time.

Pace influencing (pace influencing): by synchronizing the exerciser's pace with the music, the music tempo is accelerated/decelerated to encourage the exerciser to accelerate/decelerate in the same manner. In this case, the goal is to achieve a level of performance defined in the training program.

Different playback modes are associated with different exercise regimes, which may have different types of motion (e.g., running, cycling, interval (interval), jogging), pace, intensity, and duration. It is assumed that the user's performance is measured while exercising. User performance may be expressed as running pace in step frequency measured using a pedometer, as cycling pace in revolution frequency measured using a rotator, or as heart rate in beats per minute measured using a cardiometer. For running, the stride frequency may be taken as a unit of expression. For cycling, half of the pedaling frequency can be taken as a unit of performance. For rowing, half of the entire rowing cycle may be taken as a unit of performance. For heart rate, a single heartbeat may be taken as a unit of performance. Different units of performance may be defined for different movements. The characteristics of the music playback (e.g., tempo, volume, sound position, equalization) are then modified according to the current user performance and the currently selected playback mode. For example, the tempo of music playback may be controlled by music conversion techniques (e.g., time stretching), if desired. The mode of playback is extended to a plurality of songs, implying that a new song is selected according to the playback mode and the current user performance.

Obviously, the user controls the playback; songs and workouts may be skipped or extended using interactive controls on the player. Further, different playback modes may be selected. Finally, the duration and intensity of the current exercise is monitored for signaling the user to perform the next exercise. Songs are selected based on current user performance, current exercise goals, music playback mode, and history of songs played.

Different music playback modes require different control and music selection strategies. Further details of the standard linear playback need not be described. The system is a set-up with a portable music player, a heart rate monitor (usually comprising a chest belt with a wireless transmitter, and a receiver connected to the player) and a stride frequency monitor (using a pedometer connected to the player). The chest belt used delivers a pulse for each heart beat. The system uses a 5-point moving average of the input pulse time interval as the current heart rate. For step frequency, a 5-point moving average is also used.

A) Constant tempo: constant tempo music playback changes the tempo of a given song by time stretching if the original tempo of the song does not meet the desired tempo. In short, time stretching is the shortening or lengthening of an audio data file according to a scaling factor without modifying the pitch (pitch). Obviously, for online stretching and playback, this needs to be done faster than real-time; a time domain technique based on the synchronous overlap-and-add (SOLA) standard is employed. The scaling factor represents the ratio between the desired tempo and the original tempo of the song. Because music that is modified too much (e.g., more than 125%) sounds rather hard, time stretching cannot be done without hindrance for very large scale factors (extreme scale factors). Thus, the (next) song for playback is defined as the song whose original tempo is closest to the desired tempo and which has not been played back recently. Songs by the same artist are not played consecutively if desired.

B) Step speed matching: the exerciser is motivated by music synchronized with his movements. Pace matching synchronizes the music with the user's stride frequency. Stride frequency, or pace, is the number of times a user's foot lands in a minute, and is expressed in steps per minute (spm). The music tempo is expressed in beats per minute (bpm), where one music bar in a 4/4 beat contains 4 beats. To match the music tempo to the stride frequency, time stretching is needed to play the music faster or slower. Typical maxima for the scale factor relating to time stretching are-15% to +25%. For a song with a tempo of 100bpm this results in a stretch range of [85, 125].

Since the moment when the foot hits the ground can still coincide with a beat of music, it can also be considered a match when the pace frequency is 150spm for playing music at a tempo of 75 bpm. Note that we assume that the music tempo and the user's stride are in phase. It is generally considered that multiples (2 times, 4 times, 8 times,.. Once.)) or integral fractions (1/2,1/4,1/8.. Once.) of the step frequency match the music tempo.

For example, if the tempo of the current song is 100bpm (meaning the song stretch range is [85, 125 ]), and the stride frequency is 210spm, the system will change the music playout tempo to 105bpm. If the stride frequency is 130spm, no valid match can be found within the song stretch range. In this case, by time stretching the song, the music playout tempo is shifted to the stretched boundary closest to the stride frequency. A new song is then selected in case the current song has been played long enough (say 30 seconds to reduce repeated song changes over time). In order to make the transition without abrupt tempo changes, this new song requires a stretch range comprising the currently played tempo as well as the current stride frequency.

The playout tempo is changed in a smooth manner. A break in the music playback (hick-up) is evident when the tempo is changed immediately. On the other hand, when the change takes a long time, the system response is too slow (i.e., more than 10 seconds) to accurately match the user stride frequency. The parameter specifying the system response time in pace matching mode is T _m Which specifies the time it will take for the largest change in playout tempo as defined by the song's stretch range. A linear function is used to calculate the actual time to transition from a given playout tempo to a new playout tempo. The function is expressed by t _m Represents, it has two points, t _m (0) =0 and t _m (Δ _max )＝T _m In which Δ _max Representing the largest change in playout tempo. Linear relationships are assumed, but as a future extension, a square root relationship can be used to propagate small changes relatively quickly compared to larger changes.

When considering the above with the hypothesis T _m An instance where 1000 milliseconds, a change from 100bpm to 105bpm would require

Milliseconds.

C) The pace influence: to motivate exercisers to reach and maintain their heart rate within a certain heart rate range, the system employs a method with 4 steps.

1) Matching the music playout tempo and the user stride frequency, thereby ensuring a link between them;

2) Determining a heart rate target, and determining sub-targets of pace audio and music playout tempo from the heart rate target;

3) The change in music playout tempo is propagated from the current playout tempo to the playout tempo desired by the determined sub-goals.

4) Wait for the heart rate to stabilize.

These steps are repeated in a continuous loop until the exercise goal has been reached.

By assuming in step 1 that the music playout tempo matches the stride frequency, the playout tempo change in step 3 will have a synchronizing effect on the user stride frequency. In other words, the exerciser will automatically keep his movements in time with the later change in music playout tempo. The playout tempo and stride frequency are matched using the aforementioned pace matching pattern. Once the stride frequency is matched, the system proceeds to the next step.

In a second step, the current heart rate of the user is compared to the heart rate target of the current exercise and a difference percentage is determined. The required stride frequency is predicted by directly applying the heart rate difference percentage to the current stride frequency. From exercise physiology, we know that heart rate will vary linearly with exercise intensity. In other words, a 10% increase in heart rate means a 10% increase in stride frequency. The required step frequency is then converted into the required music playout tempo; the desired playout tempo may be a multiple or an integer fraction of the desired stride frequency.

As with the pace matching mode, when the current song cannot be stretched any further to accommodate the music playout tempo change, the system changes the playout tempo to the closest possible value (i.e., either the maximum song stretch boundary or the minimum song stretch boundary). Subsequently, a new song is selected having a stretch range including the achieved and desired music playout tempo.

The system maintains a history of heart rate measurements. Thus, it can be predicted how the heart rate (and thus the required playout tempo) will progress in the next 30 seconds. This prediction provides valuable information on what song is best selected next if the song needs to be changed for playback. This effectively minimizes the number of required song changes.

A change in music playout tempo is not immediate because a sudden change in playout tempo would result in the user not being able to keep up with the change due to ignorance or fatigue. Instead, the system considers a propagation time. As in the pace matching mode, a linear function is used to calculate the actual time to transition from a given playout tempo to a new playout tempo. This function is expressed as t _i Is shown as havingTwo points, t _i (0) =0 and t _i (Δ _max )＝T _m In which Δ _max Representing the largest change in playout tempo.

Finally, when the change has been propagated, the system waits for the user's heart rate to stabilize before making additional changes. Empirically, when testing, the user's cardiovascular system requires a period of time to accommodate the additional effort. Rather than specifying a fixed wait time after each change, the system looks for a trend of decreasing differences in consecutive heart rates with small derivative (derivative) values.

In addition, to encourage users to move more intensively, the music can be played louder, the low frequencies can be emphasized, and/or the music can be played back as if it were in front of the listener. A new song may be selected that has a tempo that corresponds to a performance level that is slightly higher than the current user performance. In addition, to encourage the user to do less strenuous, the music can be played louder, the low frequencies can be attenuated, and/or the music can be played back as if it were behind the listener. A new song with a tempo corresponding to a performance level slightly lower than the current user performance may be selected.

Feedback phase 4

The feedback stage 4 is related to offline performance storage and feedback. For user review and system learning purposes, performance data, the training program actually performed, and the list of songs actually played back are stored. The performance data may be displayed graphically for review, self-esteem, motivation, and to provide suggestions for modifying the current training method. The system obtains data for learning what music is best suited for what exercise type.

Finding music suitable for training program

The most important feature in determining the suitability of a song in a music database used in a particular exercise is its tempo. Other features include: for example, the number of taps (the number of beats in a song), the tempo, the tone, the instruments used in the recording, and so forth. Because the amount of storage space is often limited, a way to ensure that the correct music is on the portable device must be found to always be able to find the appropriate song, taking into account user preferences such as "not playing the same song twice during a training program" or "not playing songs by the same artist consecutively". This is achieved by generating a tempo probability map for each exercise and then combining these maps into an overall tempo probability map for the entire training program

In an embodiment of the method of the invention, a pace is determined which matches the exercise heart rate range in order to find music suitable for the training program. Unfortunately, this relationship is not a straightforward one. The heart rate caused by running at a certain pace varies from user to user and over time due to factors such as, among others, endurance (how long it has been running), fatigue, the health of the runner, weather conditions and time of day. However, some average tempo can be assumed, the results of which are given in the following table:

HR	sex	Age (age)	Pace speed
HR	sex	Age (age)	Pace speed	50％ 50％ 60％ 70％ 70％ 70％ 80％	For male For male For male For male Female Woman For male	Less than 30 30-40 Less than 30 Less than 30 Less than 30 30-40 Less than 30	90.2 88.6 110.1 116.5 112.9 110.7 125.0

Another important aspect of determining the average tempo for an exercise is where the exercise is located within the training program. For example, the pace required to reach heart rate 110 when starting exercise may be 90, whereas pace 80 is sufficient to reach the heart rate after 30 minutes. Thus, the fatigue curve of a user can be assumed according to gender and age, wherein the effect of time on the pace needed to reach a certain heart rate level is recorded. These rough tables and fatigue curves are based on user classification according to gender and age groups. This information can be easily entered by the user and does not require much time to perform system calibration.

A prediction of the pace for each exercise in the training program has been established, from which a gaussian distribution can be assumed, estimating the similarity of a particular tempo in the exercise to a typical heart rate range:

the mean, μ, in this simple gaussian function is the mean estimated pace for an exercise, where the standard deviation, σ, represents the width of the result plot. The standard deviation may be determined taking into account the estimated tempo of the previous exercise (since this tempo is likely to be the tempo at which the current exercise will start) and the estimated tempo of the next exercise (for similar reasons). All of the above consider statistical information about the performance of the user category (age and gender group). Furthermore, information from the training program is utilized. However, during exercise, the user's performance data is recorded (logged) and collected in a feedback phase after the exercise. Such data can be used to make the distribution more personalized.

First, the mean deviation can be determined for a certain type of exercise and can be used in combination with the statistical information that is already available to find a better standard deviation σ for such exercises. Furthermore, the average tempo μ in the function can be determined much more accurately when taking into account the user's appearance. An effective way to do this would be to generate a true curve from the user's performance and average this curve with one that has been created before. The same process can be used for user fatigue curves. Thus, the prediction of tempo is more and more accurate for each new exercise the user has completed.

A more or less exact distribution has been created for each individual exercise, and all these curves can be summarized as an overall distribution for the whole training program. The relative weight of the exercises may be taken into account. This weight is determined by looking at the duration of the exercises relative to the length of the whole training program. One of the problems that arises is that some exercise durations may be expressed in kilometers or target heart rates instead of time. For distance-sustained exercise, the distance may be multiplied by the average running speed, taking into account the user's fatigue curve. For a target heart rate, the user fatigue curve may also be taken into account.

When the overall distribution for the entire training program has been established, this can be mapped to the total playing time available on the music (player) memory, respectively. Preferably, the total playing time depends on the size of the space on the memory and on the average compression level for each song. Referring to fig. 2, this graph can be made to match one of the constraints that the song needs to satisfy in preparation phase 1.

User interface

In an embodiment of the system, the system allows the user to stop, skip, pause, or resume the workout in the training program, the songs currently playing in the playlist, and the music playback mode. The music player makes the music player itself strictly adhere to the exercise program by only playing music previously selected for the individual exercises and by taking into account the duration of these exercises or the time it takes to reach a predefined level of performance. User control of music playback is achieved by the present invention which includes a user interface system provided with access interaction controls.

The system begins by playing music that has been pre-selected for the first exercise. The exercise type is communicated to the user through speech synthesis and/or text/display prompts. Further, song information is conveyed to the user, for example, by display and/or speech synthesis. The exercise may have a predefined fixed duration or may have a variable duration. In the latter case, the workout will be extended until a predefined user performance level has been reached or as long as the predefined user performance level is maintained (see fig. 1).

When exercising and listening to music, for example, when full attention is required for other tasks such as crossing roads or talking to passersby, the user may pause music playback and exercise. The music can then be resumed and the workout resumed. If the exercise involves reaching or maintaining a particular performance level, restarting means starting the exercise again. The exercise status is reported to the user by display and/or speech synthesis.

If the particular song being played back is disliked, the user may skip the currently playing song. And then go back to the next pre-selected song. The selection of this new song may depend on the playback mode. Information about the new song is transmitted to the user. Songs may be skipped over and over until a happy song is found. The selection of new songs is cyclic, allowing the user to recover from errors and navigate through all songs pre-selected for an exercise.

When the exercise can be ended, the user is notified. Ending the exercise means that a predefined duration has expired or that the user successfully reached or maintained a certain level of performance. This information may be transmitted by an audible cue, a vibratory cue, a text/display cue, an LED light cue, or by changing the music playback characteristics (e.g., slower speed, softer). The user then decides to skip the current workout to the next workout. The user may also decide to continue with the current workout; after a short period, the device will start the current exercise again. The user can always decide to skip the current exercise even if the exercise has not yet ended. It is cyclic to proceed to the next exercise. If it is the last exercise in the training program, then the next exercise will be the first exercise in the program. This cyclic feature allows the user to recover from errors and navigate comprehensively. Other embodiments of the system may have only some of the features described above.

Referring to fig. 3, the electronic device 21 of the present invention comprises an electronic circuit 23. The electronic circuitry 23 is operative to select a subset of content items from the set of content items, the subset being smaller than the set of content items, based on feedback collected by the person in a previous physical activity, and the feedback comprising information about the previous physical activity. The electronic device 21 may be a stationary device or a portable device. For example, the electronic device 21 may be a content reproduction device, such as a portable music player, for reproducing content when a person is performing physical exercises, the electronic device 21 or a device for transferring content items to the content reproduction device or a memory used in the content reproduction device. The electronic device 21 may comprise an input 25, for example a USB port or a network connection, for receiving the subset of content items. The electronic device 21 may comprise an output 25, e.g. a USB port, a network connection, for transferring the subset of content items to the content reproduction device or a memory used in the content reproduction device. The electronic device 21 may comprise a (fixed or removable) storage means 31, such as a hard disk, a solid state memory and/or an optical disk reader. For example, the storage means 31 may store the set of content items, a subset of the content items and/or the collected feedback. The electronic device 21 may comprise rendering means 29, for example a speaker, a tactile feedback generator, a display or a light, for rendering the subset of content items. The electronic circuit 23 may be, for example, a multi-purpose processor or a dedicated application processor. Software may be used to efficiently cause the electronic circuitry 23 to perform the method of the present invention.

While the invention has been described in connection with preferred embodiments, it will be understood that modifications thereof within the principles outlined above will be evident to those skilled in the art, and thus the invention is not limited to the preferred embodiments but is intended to encompass such modifications. The invention resides in each and every novel characteristic feature and each and every combination of characteristic features. Reference signs in the claims do not limit their protective scope. Use of the verb "to comprise" and its conjugations does not exclude the presence of elements other than those stated in the claims. The use of the verb "a" or "an" preceding an element does not exclude the presence of a plurality of such elements.

'means', as will be apparent to a person skilled in the art, are meant to include any hardware (such as separate or integrated circuits or electronic elements) or software (such as programs or parts of programs) which perform in operation or are designed to perform a specified function, be it solely or in conjunction with other functions, be it in isolation or in co-operation with other elements. The invention can be implemented by means of hardware comprising several distinct elements, and by means of a suitably programmed computer. "software" is to be understood to mean any software product stored on a computer-readable medium, such as a floppy disk, downloaded via a network, such as the internet, or sold in any suitable manner.

Claims

1. An electronic device (21) comprising electronic circuitry (23), the electronic circuitry (23) being configured to:

a subset of content items is selected (6) from a set of content items, the subset being smaller than the set of content items, in dependence on feedback collected by a person (3) in a previous physical activity, and the feedback comprising information about the previous physical activity.

2. An electronic device (21) as claimed in claim 1, wherein the feedback comprises a physical state of the person (3) in the previous physical activity.

3. An electronic device (21) as claimed in claim 1, wherein the feedback comprises a pace of the person (3) in the previous physical activity.

4. An electronic device (21) as claimed in claim 1, wherein the feedback comprises a pace associated with a physical state of the person (3) in the previous physical activity.

5. An electronic device (21) as claimed in claim 1, wherein the feedback comprises information identifying which of the subset of content items have been skipped in a certain type of physical exercise.

6. An electronic device (21) as claimed in claim 1, wherein the feedback comprises information identifying an effect of one or more of the subset of content items on a physical state of a person (3).

7. An electronic device (21) as claimed in claim 1, wherein the feedback comprises a length of time of a certain type of physical exercise.

8. A method of selecting a content item, comprising the steps of:

a subset of content items is selected (6) from a set of content items on the basis of feedback collected by a person (3) in a previous physical activity, the subset being smaller than the set of content items, and the feedback comprising information about the previous physical activity.

9. Software enabling a programmable device to perform the method of claim 7.