In the year 2020, Covid has shuttered down gyms across the country and people switched to working from home and working out from home. There are two key problems I observed as I tried to workout at home.

  1. There is loads on free and high quality workout content to the point on information overload making it difficult to pick a workout.
  2. Following workout videos on phones or tablets is very inconvenient because we need to perch these devices somewhere and reach out to them whenever we need a break (say, a water break).

I figured a rich multi-media Alexa skill can solve both the problems. Alexa can quickly gather the user's requirements and recommend/play a high quality workout video that meets the users needs.

Alexa devices being communal devices are usually placed in prime spots in a user's home and so work really well for a follow along work out.

Finally, the far field voice control makes it really convenient to stop and resume the workout seamlessly.

What it does

Personal Trainer recommends and plays follow-along workouts from a curated list of high quality workout videos. Recommendations are made based on user requirements. The following workout filters are supported currently -

  1. Difficulty level (easy, medium, difficult)
  2. Muscle group (abs, legs etc.)
  3. Duration (10 minutes, 20 minutes, 1 hour etc.)
  4. Workout type (cardio, yoga, stretching etc.)

Once all the requirements are gathered, the skill will search its repo, ranks the results and plays a random video from the top results. Users will be able to control the playback using their voice or the on device touch controls.

The skill also leverages APL-A to create an immersive experience, especially at skill launch and at the end of the work out. The APL-A effects are also used to chime in a couple of times during the workout to keep the user motivated.

The skill also leverage motion sensing APIs to track the user so they can always see the screen as they move along the room while working out.

At the end of the workout, the skill provides an APL screen that lets users show their appreciation for the workout creator by liking / sharing / subscribing to their channels on Youtube.

Accomplishments that I'm proud of

  1. The skill is a very visual experience by its very nature. However, I'm proud of the fallback experience I implemented for headless devices like Echo Dot. On headless devices, we send the recommended video to the user's email address instead of playing the video.

  2. I'm proud of leveraging a whole bunch of APL technologies making seamless video playback possible on multiple devices.

What's next for Personal Trainer

  1. Recommend exercises by equipment. A user who has dumbbells should get workout videos that make use of dumbbells.

  2. Extend the skill to also provide recommendations on one exercise at a time (as against a complete workout video). For example, for a user with dumbbells, the skill can recommend bicep curls, show a short snippet and provide instructions on performing bicep curls in proper form.

  3. Keep track of estimated number of calories and provide a summary at the end of the workout.

  4. Remember user preferences and provide workout recommendations automatically for returning customers.

Share this project: