System and method for providing coaching to a user for a training session

Abstract

A computer-implemented method for generating and playing back user specific training sessions. The method comprises: • Obtaining an exercise program comprising a sequence of exercise intervals, each interval having an interval specific intensity. • Using a computer for generating a visual representation of said exercise program to be displayed on said display system, the visual representation representing the users current position in the sequence of exercise intervals in said exercise program. • Using a computer for generating speech audio based on the users position in the sequence of said exercise intervals. The speech audio is generated using an Al engine having as input: ° The interval specific intensity of the current exercise interval. ° The interval specific intensity of the following exercise interval. • Playing back said speech audio on said audio system while displaying said visual representation of the current exercise interval on said display system.

Description

[0001] System and method for providing coaching to a user for a training session

[0002] Field of the invention

[0003] The invention relates to the field of providing voice coaching of an athlete performing an exercise session using a computer system and artificial intelligence.

[0004] Background

[0005] The field of voice coaching through computer-based systems for physical activities, such as biking and running, has seen significant advancements in recent years.

[0006] T rad itional ly , coaching for these sports relied on in-person trainers who would provide real-time feedback, guidance, and encouragement. However, with the rise of mobile devices, wearable technology, and artificial intelligence, coaching systems have been developed to simulate the real-time feedback of a personal coach, making training more accessible, flexible, and data-driven.

[0007] Modern voice coaching systems utilize sensors and software to monitor metrics such as speed, distance, heart rate, cadence, and elevation. These systems obtain performance data and deliver this data as feedback through devices like smartphones, earbuds, smartwatches, or other devices. This feedback allows users to adjust their performance, thus enhancing training effectiveness.

[0008] Despite these advancements, existing systems often rely on pre-programmed routines with limited personalization. They may lack context awareness, reducing their effectiveness under varying environmental conditions like weather or terrain.

[0009] Additionally, many systems do not support adaptive training protocols that respond dynamically to real-time biometrics and environmental data, limiting their utility for users seeking responsive and individualized training programs and coaching.

[0010] An example of a system providing voice coaching is the system described in the patent application EP4238619A1 , where images of body positions are used by an Al to provide voice coaching to the user.

[0011] A summary of the invention

[0012] A first aspect of the present disclosure is a computer-implemented method for generating and playing back user specific training sessions on a computer controlled training system comprising a display system and an audio system. The method comprises:

[0013] • Obtaining an exercise program comprising a sequence of exercise intervals, each interval having an interval specific intensity.

[0014] • Using a computer for generating a visual representation of the exercise program to be displayed on the display system, the visual representation representing the users current position in the sequence of exercise intervals in the exercise program.

[0015] • Using a computer for generating speech audio based on the users position in the sequence of the exercise intervals. The speech audio is generated using an Al engine having as input: o the interval specific intensity of the current exercise interval, o the interval specific intensity of the following exercise interval,

[0016] • Playing back the speech audio on the audio system while displaying the visual representation of the current exercise interval on the display system.

[0017] An alternative to the first aspect of this disclosure is a computer-implemented method for generating and playing back user specific training sessions on a computer- controlled training system comprising a display system and an audio system, wherein said method comprises:

[0018] • obtaining an exercise program comprising a sequence of exercise intervals, each interval having an interval specific intensity;

[0019] • using a computer for generating a visual representation of said exercise program to be displayed on said display system, the visual representation representing the user’s current position in the sequence of exercise intervals in said exercise program;

[0020] • using a computer for generating speech audio based on the user’s position in the sequence of said exercise intervals, wherein said speech audio is generated using an Al engine having as input: o the interval specific intensity of the current exercise interval, o the interval specific intensity of the following exercise interval,

[0021] • playing back said speech audio on said audio system as forward-looking voice coaching based on the current and following exercise interval including the intensity of the current and following exercise interval, while displaying said visual representation of the current exercise interval on said display system. An exercise program comprises multiple intervals each with a specific intensity. Such a program may be designed to improve certain skills of a user such as endurance, sprint, breathing, mentality or any other important skills for doing sports. Training programs may be designed generically to fit every person. The effect obtained by the user / athlete of such a program may vary a lot as people are different and have different motivation. Therefore, an exercise program may advantageously be customized to each person. This may be done in multiple ways. An example is to have the person inputting a goal and the amount of time available for training and the type of training e.g. biking. A computer may use an artificial intelligence to create an exercise program based on all exercise programs in the system which may include previous exercise programs from the specific user / athlete. Thus, a personalized exercise program for providing the most effective training can be obtained and used for the method described above.

[0022] A visual representation of an exercise program may be any visual representation e.g. a graph illustrating the intensity of each interval, a number representing the intensity, and a number representing the length of the interval and time to the end of the interval, or a virtual world having indicators of where the interval ends and starts and the intensity may be illustrated by the inclination of the road or the type of terrain. Having a visual representation of the exercise program provides information of the intervals’ intensities, the lengths of the intervals, and the remaining time of the current interval, thus, lots of information of the program to the user during performance of the exercise program. The user can then use the information when controlling the training system e.g. an exercise machine like a bike or a treadmill or other training equipment. The user may adapt the remaining power to fulfill the exercise program such that e.g. no energy of the user is left when the program has ended. The user may also time the energy such that a specific energy level is left in the end of the exercise program. Thus, the user uses this information to control the training system accordingly.

[0023] A speech audio is also created and this speech audio is outputted to the user such that the user is encouraged / coached to follow the exercise program. The exercise program is also illustrated visually through the visual representation. Thus the control of the training system may be even more improved by outputting audio coaching to the user during the exercise. One of the effects of using this method is to enhance the control of the training system that the user is using for the training.

[0024] The speech audio may be played back on an audio system as forward-looking voice coaching as the speech audio has been created having the intensities of the current and next interval as input. Thus, the forward-looking voice coaching can coach accordingly to what the user is doing right now and what is coming up in the training program or exercise program. The exercise program may be fixed from the start of the exercise session and thus the forward-looking voice coaching is adapting to the exercise program and the intervals thereof. The forward-looking voice coaching may adapt to the intervals going through the exercise program as when the user arrives to the next interval, the Al engine has as input the interval after the current interval such that the forward-looking voice coaching is obtained. The data of the intervals may be dynamically loaded as inputs to the Al engine as the exercise program is performed.

[0025] In embodiments, the training system comprises a training machine. The training machine may be a home trainer bike, a treadmill, a rowing machine, or any other suitable training machine.

[0026] In embodiments, the input to the Al engine comprises the length in time of the intervals for which the intensity is provided.

[0027] Hereby the Al engine may use this information to provide more precise speech audio such that the user may be guided to control the exercise machine with more precision.

[0028] In embodiments, the input to the Al engine comprises the interval specific intensity of the previous exercise interval.

[0029] Hereby the Al engine may take into account what the user have just performed in the exercise program, such that the user is not encouraged to get going if a high intensity interval has just been performed and so on.

[0030] In embodiments, the input to the Al engine comprises the interval specific intensity of all of the intervals of the exercise program. Having all the intervals included as inputs to the Al engine the speech audio may provide more realistic outputs and comments to the user during the exercising for the user to control the training system according to the exercise program.

[0031] In embodiments, the input to the Al engine comprises interval data from one or more previous exercise sessions.

[0032] The Al engine may further take into account how the user was coached through the speech audio in the previous sessions. This is one of the advantages of using an Al engine as lots of data can be processed fast to provide speech audio personalized to the user.

[0033] In embodiments, the speech audio is transformed by an Al engine to imitate a specific voice before outputted by the audio system.

[0034] Transforming the speech audio provides the possibility to output the speech audio as any voice that the Al engine can transform the speech audio to. This may include any specific person, a fictious person, a friendly trainer, a military commander or a family member. For family members and such special persons, a number of recordings may need to be inputted to Al for the Al to imitate e.g. an uncle, a father or a mother. In this way, the Al engine may be trained for imitating almost any person.

[0035] In embodiments, the training system comprises an intensity sensor.

[0036] In embodiments, the method comprises a step of continuously obtaining measurements from the intensity sensor related to the intensity performance of the user, the measurements being comprised in the input to the Al engine.

[0037] Having an intensity sensor and continuously obtaining measurements of the intensity of the user provides inputs to the Al engine which include information on the current intensity of the user and previous intensities of the user. This information may be used to guide the user through the speech audio to follow the exercise program even closer than the user did without this feedback from the intensity sensor. In this way the speech audio may adapt based on the intensity measurements continuously. The Al engine may from the measurements determine if the user has a lower intensity than the interval being performed or a higher intensity and thus output speech audio that provide this information to the user such that the user may adapt the training and control of the training system accordingly. Thus, the user may interact with the computer and the Al engine by providing input via the intensity sensor and adapt the training and the control of the training system according to the instructions received through the audio system from the computer. This way the user interacts with the computer and the speech audio in order to control the training system such that the user intensity matches the intensity of the exercise program which enables the user to follow the exercise program more effectively and precisely than without this interaction.

[0038] The exercise program is normally not adapted during performance of the exercise. The audio speech is adapted in order to guide the user through the exercise program with as much control of the training system as possible. This system may provide the best possible coaching of a user for a given exercise program and providing dynamic forward-looking voice coaching to the user as the user is controlling the training system, which may also be denoted the exercise system.

[0039] In embodiments, the method comprises generating a virtual training world based on the exercise program, and wherein the visual representation comprises a visual representation of the virtual training world.

[0040] By providing a virtual world as part of the visual representation, the user may get an experience that the training is a game which could make the training easier for the user to complete and the exercise program may be easier to interpret for the user such that the training system may be controlled accordingly.

[0041] A second aspect of the present disclosure is a training system configured to performing the method or part of the method as described above in the first aspect of the invention, wherein the training system comprising:

[0042] - a display system;

[0043] - an audio system; and

[0044] - a processing unit configured to run an Al engine.

[0045] By having a system which is able to perform the method according to the first aspect of the invention the same advantages as the performing the method is obtained by this system. It provides a system able to voice coach the user to follow the exercise program and control an exercise machine according to the program in a more precise and consistent manner than without the voice and visual coaching.

[0046] In embodiments, the training system comprises an intensity sensor.

[0047] By comprising an intensity sensor, the system is able to obtain intensity data of the user continuously and send this as input to the computer and the Ai engine. Hereby the intensity information of the user can be used to voice coach the user.

[0048] In embodiments, the training system comprises an exercise machine.

[0049] In embodiments, the exercise machine is a home trainer bike.

[0050] By including a bike in the training system, the system is able to provide voice coaching specifically to cyclists as this is a training form where coaching is used a lot and where the personal coach could be omitted and replaced by this training system.

[0051] Summary of figures

[0052] Fig. 1 illustrates a training system and method.

[0053] Fig. 2 illustrates a training system and method with feedback from a sensor.

[0054] Fig. 3 illustrates an visual representation of an exercise program.

[0055] Fig. 4 illustrates a computer-implemented method for generating and playing back user specific training sessions on a computer controlled training system.

[0056] Fig. 5 illustrates a training system and method where the visual representation includes a representation of a virtual world.

[0057] Detailed description of the invention

[0058] A training system 10 for voice coaching an athlete for performing a training program or exercise program 14 is disclosed and illustrated in Fig. 1. The system / training system 10 uses artificial intelligence to create outputs from which speech is generated and outputted through an audio system 20 to the athlete performing the training program.

[0059] A training program / exercise program 14 comprises multiple intervals where each interval is defined by having a specific intensity. An intensity may be defined as any parameter describing an intensity. The parameters may be Pulse / Heart Rate, Watt, FTP, %FTP, Position, Speed, Distance Traveled, MAP, %MAP, FTHR, %FTHR, Power Zones, LTHR, %LTHR, VO2max, %VO2max, RPE Levels, Kcal, Strokes, Time (time will affect the RPE levels). These are just examples of data that can determine intensity in an exercise program 14 and it will depend on the exercise activity, which could be anything from cycling, running, skiing, rowing or the like where it is possible to train with different intensities or any other exercise where intervals are used for training.

[0060] An interval is defined by a length in time and the intensity for which it is to be performed. The length in time of an interval may be defined by a start time and an end time.

[0061] By providing an exercise program 14 data of at least one intensity parameter is provided together with the time interval of the exercise program 14 for which this intensity is to be performed by an athlete, this intensity data for the exercise program 14 is denoted the program data. An exercise program 14 comprises at least two intervals and therefore at least two intensities.

[0062] The program data comprises the current interval and the following interval. Hereby the program data may change during performance of the exercise program 14 such that the data continuously updates during the training.

[0063] The program data may comprise data of the previous interval.

[0064] The program data may comprise data of at least two, three or four following intervals. Preferably the program data may comprise data of all intervals in the exercise program 14 including the intensities and the length of each interval.

[0065] The program data may include data from previous training sessions and exercise programs 14 and the time these earlier training sessions were performed. Thereby the Al engine may also use this information to provide coaching as the user may be tired or energetic from the earlier training and potentially the Al engine knows how the last training went and what the intensity and length of the training was.

[0066] As illustrated in Fig. 1 and Fig. 3 the training system 10 comprises a display for displaying a visual representation 16 of the exercise program 14. The display 12 may be any display 12 like a tv screen, a laptop computer, the display 12 of a smartphone, or another suitable display.

[0067] The exercise program 14 is visually represented and displayed on a display 12 for the user to be able obtain information of the timely position in the exercise program 14 during performance of the exercise program 14. Thus, the user may visually see what the next interval of the exercise program 14 is and prepare and react accordingly to this information. The user may also be able to see multiple or all the remaining intervals of the exercise program 14. The display 12 may display all intervals of the exercise program 14 for the user / athlete to know the progression of the exercise program 14 no matter which interval the athlete / user 24 is performing currently.

[0068] The user performing intervals of an exercise program 14 may improve the experience when the user can see and mentally prepare for the challenges of intervals coming up in the exercise program 14. By having the information, the user may better follow the exercise program 14 and control the exercise system / training system according to the program. The user may also have a more satisfactory experience of the training than without this information.

[0069] The visual representation 16 of the exercise program 14 may include a visual representation 16 of a virtual training world generated by a computer 23 as illustrated in Fig. 5. The virtual world 26 is generated based on the exercise program and may indicate when the intervals are completed and / or start. The visual representation 16 may be included in the generated virtual world 26 or displayed separately from the virtual world 26. The virtual world 26 could be a cartoon-like underwater world, a world in space or any other world based on the creativity and fantasy of the virtual world 26 designer. It is all about designing a world that enables illustration of speed and inclination and at the same time which motivates the participant. Another example is that the virtual world 26 is a known world from cartoons, movies, computer games or the like, where elements being passed by the participant during the exercise correspond to elements and environments which can be recognized from these well- known worlds. As an alternative to the fictive environments, the world could also be completely or at least partially based on actual real-world data e.g. from google maps or similar. Thereby the inclination and speed adapt to the intensity data and the inclination of the surface from the real world would adapt to the intensity. A result would be that Alpe d’Huez could be illustrated as flat if the participant has a low intensity when passing the area being a mountain in the real world, or when passing a real-world flat area with a high intensity it could be presented with an inclined surface. Another possibility is that the speed is reduced when passing a mountain like Alpe d’Huez with low intensity, and the speed then increases when an interval with higher intensity is performed. The participant might experience passing houses from the home city of the participant. The virtual training world may be generated continuously as the user is training and going through intervals of an exercise program 14. The virtual world 26 may be generated such that it fits intervals of a training program e.g. by changing the look of the virtual world 26 when a new interval is entered or marking the end and start of the intervals by a marker or highlighter.

[0070] The display is connected to a computer comprising a processing unit. The computer may be a computer of a server system, and the connection may be any data connection like through any internet connection. The computer may be a local computer. This processing unit 22 is used to generate the visual representation 16 of the exercise program 14.

[0071] The training system 10 comprises an audio system 20 being able to output speech audio generated by the processing unit 22. The speech audio may be generated directly by the Al engine or generated as text by the Al engine and subsequently output as audio in the form of speech, talk, or shouting to provide voice coaching of the athlete / user 24. Thus, the speech audio may be generated in multiple ways. The audio system 20 may comprise at least one speaker. The audio system is in communication with the processing unit 22 to output speech audio when a program of the processing unit 22 instructs an audio output 21 .

[0072] The speech audio is generated using a computer / processing unit 22 using an artificial intelligence also denoted an Al engine. The speech audio may be described as audio data that may be outputted as audio on an audio system. The speech audio may be transformed by a computer before being outputted on the audio system. The artificial intelligence may comprise any standard machine learning algorithm suitable for generating audio or text. The Al engine is used to generate speech audio. The inputs to the Al engine contain the information of at least the intensity of the current interval and the next interval in time. This data is also denoted the program data. This information given as input to the Al engine ensures that the output generated takes into account the next interval such that the audio speech / voice coaching ensures that the user is able to perform the next interval and is performing the current interval with the desired intensity or voice coaching. The program data may as earlier described comprise information of the previous interval and / or information of more of the intervals of the exercise program 14. By providing more information of the intervals of the exercise program 14, the Al engine may provide coaching that takes into account what is coming up in the exercise program 14 as the user / athlete is performing the program. The Al engine may also take into account that the user may just have performed a very intense interval and needs to get the breathing under control even though this interval should be easy. Thus, by inputting the program data to the Al engine, the output will coach the user / athlete to follow the exercise program 14 even better than if the Al just had the intensity of the current interval as input this ensures that the user can control the training system according to the exercise program by following the voice coaching. The program data provided as input to the Al engine may additionally comprise the length of the intervals. E.g. the current interval may have a length of 1 min and the computer will during the exercise be able to calculate that the user is 20 seconds into the interval and thus that 40 seconds remain of the interval. Thus, the time data may comprise both the length of the interval and alternatively also the timely position of the user performing the exercise program.

[0073] The generation of speech audio when generated as text that is converted into audio may include using a second artificial intelligence to generate the voice talking text generated by the first Al engine. The second artificial intelligence may generate speech such that the voice of specific persons can be imitated to coach the athlete / user 24 even though the specific person has not been coaching anyone to perform an exercise program 14. This may be done by training the second artificial intelligence with audio data from the specific person such that the artificial intelligence can learn how the person would talk or speak. Using an artificial intelligence to generate a voice or audio speech provide the advantage that the system may be voice coaching with a lot of different voices or sounds, and the system is thereby very flexible to the preferences of the user. This feature may correspond to selecting a coach which fits the user and may be personal, and thus it is an advantage to be able to shape the coach voice to any desired voice.

[0074] The generation of speech audio may be performed by a single artificial intelligence that takes the program data as input and provide audio as output. The speech audio may describe forward-looking voice coaching as the audio speech is created with information about what interval is coming up in the exercise program and the coaching takes this into account. Thus, the exercise program itself will normally not change or adapt during the exercise, but the user activity and intensity will adapt based on the forward-looking voice coaching. Thus, the coaching may adapt during an interval of the exercise program and the user’s intensity may change or vary during an interval of the exercise program.

[0075] One of the objects of the system is to increase at least one of the user’s / athlete’s performance, gain, mood and motivation during and after exercising.

[0076] The training system may comprise a voice coaching subsystem configured to generate and deliver forward-looking auditory guidance to the user during execution of the exercise program. The voice coaching subsystem operates in conjunction with the display system and may be executed by a computer system comprising one or more processors.

[0077] In operation, the computer receives the exercise program, which is defined as a sequence of exercise intervals, each interval having an associated interval-specific intensity parameter. As the user progresses through the exercise program, the computer may continuously determine the current interval position within the sequence and identifies the subsequent interval that immediately follows the current interval. To provide meaningful auditory guidance, the computer utilizes an Al-based speech generation engine configured to synthesize speech audio such as from structured exercise-related input data. The input to the Al engine comprises at least (i) the interval-specific intensity of the current interval and (ii) the interval-specific intensity of the following interval. Based on these inputs, the Al engine generates voice coaching that reflects both the user's present workload and the anticipated workload transition in the upcoming interval. This enables the system to produce forward-looking coaching prompts that prepare the user for changes in intensity before such changes occur. The speech audio generated by the Al engine may include, but is not limited to, motivational cues, pacing instructions, warnings of upcoming intensity increases, recovery guidance during intensity decreases, and preparatory coaching messages corresponding to transitions between different interval types. By incorporating information relating to both the present and the upcoming intervals, the system delivers coaching that is temporally anticipatory rather than reactive, thereby allowing the user to adjust performance in advance of interval transitions. The computer may output the generated speech audio to the audio system, which may include loudspeakers, headphones, or any suitable audio playback device. The playback is synchronized with the visual representation of the exercise program shown on the display system, such that the user simultaneously receives graphical information regarding the current interval and auditory information preparing the user for the following interval. The combined presentation results in a coherent and intuitive training experience wherein the user receives real-time, predictive voice coaching aligned with the structure and intensity progression of the exercise program.

[0078] In some embodiments, the Al engine may further adapt the tone, timing, and content of the voice coaching based on user-specific preferences or past training behavior, thereby personalizing the auditory feedback for enhanced training effectiveness.

[0079] The voice coaching may be both forward-looking and backward-looking as the speech audio may be created based on data for both upcoming intervals, and intervals that the user has performed.

[0080] The generated speech audio is played back to the user / athlete on the audio system 20 and thus providing live coaching based on the program data as illustrated with the full lined arrow in Fig. 1 and Fig. 2. This way the user gets voice coaching without having a personal trainer standing next to them. Thus, the user may be encouraged to control the exercise equipment according to the exercise program 14. This improves the user experience and the effect of the training for the user to get the wanted effect by following a specific exercise program 14 even more precisely than without a voice coach.

[0081] The following computer implemented method may be performed having an exercise system. The method is illustrated in Fig. 4. As described in detail above the method comprises the steps of:

[0082] 200 Obtaining an exercise program comprising a sequence of exercise intervals, each interval having an interval specific intensity.

[0083] 202 using a computer for generating a visual representation 16 of said exercise program 14 to be displayed on said display 12 system, the visual representation 16 representing the user’s current position in the sequence of exercise intervals in said exercise program 14; 204 using a computer for generating speech audio based on the user’s position in the sequence of said exercise intervals, wherein said speech is generated using an Al engine having as input: o the interval specific intensity of the current exercise interval, o the interval specific intensity of the following exercise interval,

[0084] 206 playing back said speech audio on said audio system 20 while displaying said visual representation 16 of the current exercise interval on said display 12 system.

[0085] The training system 10 may comprise one or more intensity sensors capable of measuring a parameter describing the intensity of the athlete during activity. The intensity sensor may be a power meter, a heart rate sensor, a pace / speed sensor, an accelerometer, a cadence sensor, oxygen saturation sensor, muscle oxygen sensor, or any other sensor for measuring the intensity of a performance of an athlete.

[0086] The intensity sensor may be a separate sensor, or a sensor integrated in a smart watch or another device. The sensor may be part of the exercise machine that the athlete is using to perform the training. This may be a home trainer exercise bike, a treadmill, a rowing machine, a skiing machine, a bike, a running device, or another training devices.

[0087] The intensity sensor, also denoted the sensor, is communicatively connected to the processing unit 22 or to a storage connected to the processing unit 22 such that the sensor data can be transmitted to the processing unit 22 as the measurements are completed. This sensor output 28 of the intensity sensor may optionally be sent to the processing unit 22 and be inputted to the Al engine as illustrated in Fig. 1 . In Fig. 2 this sensor output 28 is illustrated as part of the system and thereby not being optional for the embodiment illustrated in Fig. 2.

[0088] The intensity sensor may collect sensor data of the intensity parameter of the user / athlete and thus represent what intensity level the user is performing at in real time. Measurements may be performed at regular intervals such as every 20 milliseconds, 50 milliseconds, 100 milliseconds, 500 milliseconds, every second, or every two seconds. The measurement may be an average of the user’s performance in a time interval. The training system 10 may provide live feedback to the athlete by obtaining measurements of an intensity parameter of the athlete / user 24 and input the sensor data obtained to the Al engine. Now the data inputted to the Al engine is the program data and the sensor data. Thus, the Al engine may compare the performance of the user / athlete to the intensity level of the current interval. Together with the information of the following interval the Al engine may provide speech audio comprising information of sensor data such that the user knows at what intensity is being performed currently and the intensity the current interval. Thus, the users do not need to compare these values themselves as the training system 10 performs a comparison and output if the user needs to keep the intensity or increase or decrease the intensity in this interval. The measurements obtained by the intensity sensor may be displayed on the display such that the user can follow the performed intensity directly. The measurements may also be only provided to the Al engine to enhance the interaction between the speech audio and the user as the speech audio can be adapted according to the intensity measurements continuously during the exercise program.

[0089] The training system 10 can by having sensor data and program data provide feedback to the user / athlete by visually representing intervals of the exercise program 14 and provide speech audio based on the sensor data and program data such that the user / athlete can adjust the training accordingly. In this way, the user can have a continuous interaction with the training system 10 where the speech audio outputted to the user is generated continuously informing the user at what intensity the user is performing at currently. Thus, the user can interact with the training system 10 and the Al engine by adjusting the intensity depending on the speech audio received live. In this configuration the exercise machine is controlled by the user and the controlling is enhanced by the interaction between the user and the training system 10 through the speech audio.

[0090] Here is a list of data that may be inputted to the Al engine as data to provide a well coaching system increasing the user control of the training system:

[0091] Intensity of the current interval. Intensity of the following interval. Intensity of all intervals of the exercise program.

[0092] - Timely length of the current interval.

[0093] - Timely length of the following interval.

[0094] - Timely length of all intervals of the exercise program. Current point in time of the exercise program.

[0095] - Type of training machine. Sensor type of training system.

[0096] - Training data of previously exercises performed by the user. Current intensity performed by the user as measured by the sensor.

[0097] Data that may be outputted by the speech audio or displayed at the display: Intensity of the current interval.

[0098] Intensity of the following interval.

[0099] - The intensities for all intervals of the exercise program.

[0100] - Total time of the exercise program.

[0101] Current point in time of the exercise program.

[0102] - Time left of the exercise program.

[0103] - Time next interval of the exercise program.

[0104] Number of intervals left in the exercise program.

[0105] Outputs of the intensity sensor.

[0106] Difference between the intensity measured by the intensity sensor and the intensity of the current interval.

[0107] The training system 10 may be made for any sport / exercise such as cycling, running, rowing, and so on. Preferably the training system 10 is made for cyclists and one of the parameters of an athlete that describe the intensity of the athlete during cycling.

[0108] Below is a list of reference signs used in the detailed description of the present disclosure and in the drawings referred to in the detailed description of the present disclosure.

[0109] 10 Training system

[0110] 12 Display

[0111] 14 Exercise program

[0112] 16 Visual representation

[0113] 18 Exercise device

[0114] 20 Audio system

[0115] 21 Audio output

[0116] 22 Processing unit Computer Athlete / user Virtual world Sensor output Obtaining an exercise program Generating a visual representation of the visual representation Generating speech audio playing back the speech audio on the audio system

Claims

1. CLAIMS1 . A computer-implemented method for generating and playing back user specific training sessions on a computer-controlled training system comprising a display system and an audio system, wherein said method comprises:• obtaining an exercise program comprising a sequence of exercise intervals, each interval having an interval specific intensity;• using a computer for generating a visual representation of said exercise program to be displayed on said display system, the visual representation representing the user’s current position in the sequence of exercise intervals in said exercise program;• using a computer for generating speech audio based on the user’s position in the sequence of said exercise intervals, wherein said speech audio is generated using an Al engine having as input: o the interval specific intensity of the current exercise interval, o the interval specific intensity of the following exercise interval,• playing back said speech audio on said audio system as forward-looking voice coaching based on the current and following exercise interval including the intensity of the current and following exercise interval, while displaying said visual representation of the current exercise interval on said display system.

2. A computer-implemented method according to claim 1 , wherein said input to said Al engine comprises the length in time of said intervals for which said intensity is provided.

3. A computer-implemented method according to any of the preceding claims, wherein said input to said Al engine comprises the interval specific intensity of the previous exercise interval.

4. A computer-implemented method according to any of the preceding claims, wherein said input to said Al engine comprises the interval specific intensity of all of said intervals of said exercise program.

5. A computer-implemented method according to any of the preceding claims, wherein said input to said Al engine comprises interval data from one or more previous exercise sessions.

6. A computer-implemented method according to any of the preceding claims, wherein said training system comprises an intensity sensor.

7. A computer-implemented method according to claim 6, wherein said method comprises a step of continuously obtaining measurements from the intensity sensor related to the intensity performance of the user, said measurements being comprised in the input to the Al engine and the voice coaching is adapted based on said measurements, wherein the user control of the training machine is increased and the user intensity may be adapted based on the forward-looking voice coaching.

8. A computer-implemented method according to any of the preceding claims, wherein the method comprises generating a virtual training world based on said exercise program, and wherein said visual representation comprises a visual representation of said virtual training world.

9. A training system configured to performing the method of any of claims 1-8, wherein the training system comprising:- a display system;- an audio system; and- a processing unit configured to run an Al engine.

10. A training system according to claim 9, wherein the training system comprises an intensity sensor.

Images