Request for another learning feature

Synthesia is a living project. You can help by sharing your ideas.
Search the forum before posting your idea. :D

No explicit, hateful, or hurtful language. Nothing illegal.
Post Reply
Posts: 2

Post by Badabing » 04-09-18 6:01 am


I've been using Syntesia for a month now and I'm really pleased by how fun and easy it is. However there is one thing that bothers me a bit. The step from the first learning feature to the next one seems really steep. You can do a lot of customization yourself to make the transition more smooth, like reducing the tempo and setting up loops. I do however feel that the experience would be a lot more enjoyable if those things were automated.

Something like this:

Synthesia cuts the song into passages (5 seconds or so). When it finishes the first passage at 100% speed it checks whether you've reached a certain score. If you haven't, it repeats the same passage at a lower speed. If you fail to reach the certain score again, it repeats the same passage at a lower speed again until you succeed. Whenever you succeed it speeds up again while repeating the same passage until you hit 100% speed and hit the right score. It then moves on. EDIT: or maybe it should stay at same speed a couple of times before going up in tempo so that you'll not get too confused by constant change in tempo. Maybe it should require you to hit the score 2 or 3 times (or more) in a row before it speeds up.

This would feel like how you'd learn to play on your own and also by reducing the tempo for you it would keep reminding you to relax your muscles and not tense up when you reach difficult parts of the song.

I don't know how hard this would be to implement but for me it would be a huge deal.

Posts: 58

Post by cmplays » 04-10-18 12:40 am

It sounds like a great quality of life feature in theory, but I don't think it's that easy to implement. First of all, how you split up the song depends on its difficulty. If you have only 5 notes in 5 seconds, it's one thing. If you have 25 notes in the same interval, it's a whole other level of difficulty.

But the real issue (at least the way it's been for me) is that even if you master a segment at 100% speed, it doesn't mean that you've really mastered it. First of all, the transition for this segment to the neighboring segment has to be practiced independently. So your next segment to practice should not be an adjacent segment, but rather an overlapping segment. But the most bizarre twist is that even if you play a segment perfectly on its own, it doesn't mean that you will play it perfectly in its context. There is a lot of interference from muscle memory built up from what you play before it, plus your mind might also still be dwelling on what you were playing must before, and not be ready to transition. Which means that in order to really master it, you have to play not just the segment, but that segment plus segments before and after, and than the segments before and after the larger segment, and so on in ever widening chunks.

An implementation that would take this into account is doable of course, but definitely not simple.

Posts: 2

Post by Badabing » 04-10-18 6:55 am

Good point. Didn't think of that. Maybe it could cut the song up differently after each play-through (by pushing the segments forward or backwards for instance)? But definitely more complicated and not as clear cut as I initially thought.

Posts: 27

Post by miek » 04-11-18 4:03 pm

I agree with all of that.

Maybe to simplify it for the software, some feedback about what kind of quality (speed, correct notes and lengths etc) you are getting and how it varies within the song would be the easiest way of implimenting it. What we get now is for the whole song, but how this might work is, after the song is over, as well as the existing overall feedback, this is broken down into bars maybe as a graph or if not maybe a table of numbers. From this you could see which bits you are slower or less accurate on, and maybe the user put more emphisis on practicing that part of the song. I don't think it would have to identify exact issues within each bar, more of an average showing those parts of the song you could practice more.

Some of this information is there already - Where it shows the time taken (the extra time) in loop mode. If I was to keep an eye on this, it would only increase if I was slower than the song, and the extra notes part is the same. So could this be the basis for the extra feedback that would be more helpful, not necessarly a big change?

I know on my favourite songs that I practice I can get eg. 50% on "actual speed" but some parts must be played too fast and some too slow. I think I need to practice the difficult parts more - if Synthesia could help show these parts it would be even greater than it is now :D

Posts: 12196

Post by Nicholas » 04-14-18 2:32 am

miek wrote:... after the song is over, as well as the existing overall feedback, this is broken down into bars maybe as a graph or if not maybe a table of numbers.
This will be precisely the focus of the Synthesia 12 release! :D

The goal is to completely revamp the way Synthesia tracks progress down to the per-measure level. It won't matter anymore if you practice the entire song all the way through. Any amount of playing will record data for the measures you attempted. (If you play a short loop many times in a row, those measures will receive lots of "evidence", etc.)

Posts: 27

Post by miek » 04-15-18 5:57 am

I can't wait for Synthesia 12 :D But the current Synthesia is still great!

Post Reply