2010 Roadmap

Have questions? Just saying hello? This is the place.
No explicit, hateful, or hurtful language. Nothing illegal.
Nicholas
Posts: 13137

Post by Nicholas »

You people don't skip a beat: Practice Mode. Practice Mode Scoring. Bookmarks. Looping. Looping (again). Looping (better). Looping (for real this time). And now fingering. Turn-around time of like one day. :D

Let's talk about fingering.

You want manual fingering. Cairnz wants some sort of fingering import.

What will it take to get it crossed off the list?

My main assertion is still that unless you're manually entering from some other piece of reference material (e.g., a book or something), the person that wants the fingerings the most is in the weakest position to provide them by themselves. So to me, manual entry is of fairly limited value.
vicentefer31
Posts: 899

Post by vicentefer31 »

Nicholas wrote:You want manual fingering.
What I wanted when I did the collage it was do something "funny" . I thought when Nicholas or others users see this new collage they will think "Oh, no, Vicentefer31 again!!!" :lol:
Picasso: I am always doing that which I cannot do, in order that I may learn how to do it.
Nicholas
Posts: 13137

Post by Nicholas »

vicentefer31 wrote:... they will think "Oh, no, Vicentefer31 again!!!" :lol:
I think that every time I see you've made a post. ;)
Raymond
Posts: 528

Post by Raymond »

I think you should wait until you have the online scoring, so people can get the fingering data from the web. Or have it auto download when it gets the online scores.
vicentefer31
Posts: 899

Post by vicentefer31 »

Nicholas wrote:
vicentefer31 wrote:... they will think "Oh, no, Vicentefer31 again!!!" :lol:
I think that every time I see you've made a post. ;)
I knew it ;)
Picasso: I am always doing that which I cannot do, in order that I may learn how to do it.
Nicholas
Posts: 13137

Post by Nicholas »

Raymond wrote:I think you should wait until you have the online scoring...
Online scoreboards are definitely next. Like... very next. Next month. It's actually a pretty huge task, and I want it live before Oct-26 with a little time left over to iterate on it.
Raymond
Posts: 528

Post by Raymond »

Do you think the fingring could be included with the online scoring system?

It would make it way easyer for people to get fingering info.
User avatar
cairnz
Posts: 182

Post by cairnz »

HAha awesome. <- in reply to the gps image.

Didn't TonE provide some fingering information for the Czerny/Hanon exercises?

I think a generic system to allow reading fingering information would be sufficient. It's mostly for practicing though, Synthesia is not so much a game for me (yet) - but more the best way for me to learn music and playing the piano.

(This brings on a whole 'nother bunch of features I'd like implemented: ear training (synthesia sends a note to midi, you playback - will provide a better layout for this when i have time to write up a proper specification), chord library, metronome exercises (hold the speed constant), and so forth.

I don't know exactly how fingering information is to be provided into Synthesia because i don't know enough about midi structure to see how it'd fit in. But from what i can imagine, you would automatically scan the folders for .mid files and .fin (or .txt or whatever) files - same name equals fingering information for this midi. In this file it specifies which track it belongs to, and then just adds fingering information for every note in that track. Having possibility to write this in plaintext would be beneficial. I think the post about automatic fingering prediction has it pretty much covered: viewtopic.php?f=5&t=985
Electrode
Posts: 195

Post by Electrode »

Nicholas wrote: Let's talk about fingering.

You want manual fingering. Cairnz wants some sort of fingering import.

What will it take to get it crossed off the list?

My main assertion is still that unless you're manually entering from some other piece of reference material (e.g., a book or something), the person that wants the fingerings the most is in the weakest position to provide them by themselves. So to me, manual entry is of fairly limited value.
Not necessarily. If people ask for fingering help on the forums, they're going to need a way to input those fingerings. If fingerings were downloadable online and sent to Synthesia, people would need a way of editing them to fit their own hands and finger shapes. What's good for the goose may not be good for the gander.

In my experience, algorithms are always a bit ropey, because they simply cannot take into account all possible combinations. The more variables you provide to tweak the output, the more complex the algorithm becomes, and the less useful it becomes for the casual user.

I have tried Midi Illustrator's fingering algorithm (bought the program mainly for it, as well as the notation capabilities to practice proper sight reading) and it's underwhelming. The best way to use it is still to work out some of the fingering yourself before getting it to generate the rest. But then again, if you do that, chances are that the rest of your fingers will already be in the right places to hit the notes you don't need to put fingering down for, making the algorithm pretty useless. Even if you do leave it to automatically generate everything, you will still need to go back and edit anyway, which would take you the almost the same amount of time as figuring out all of the fingering by yourself. Blindly taking any given fingering as gospel is a dangerous thing, because it can handicap you and possibly cause physical injury. You must always be prepared to tweak things to suit your own purposes. Anyway, I digressed from the subject a little, so to get back on track...

If you leave the algorithm to figure out fingerings by itself, the quality is largely dependent on the length of the passage you are trying to get fingering for. Longer passages can be incredibly weird-looking. Feeding it chunks of two bars is alright, depending on the material, but give it eight bars (of anything) and it falls to pieces unless you help it out. Leaps are also hit-and-miss (mostly miss unless helped).

My piano teacher says that everyone should eventually be able to write their own fingerings for a passage, based on what they personally can accomplish, and what they feel comfortable with - but this obviously goes hand-in-hand with music theory knowledge (knowledge of scale/arpeggio fingering etc.). Nobody can put down better fingerings for you than yourself, after all.

Given my experience with Midi Illustrator, I would never use an automatic fingering generator on its own, and my reasoning is that if I need to give it help, I might as well do the whole thing myself. Manual fingering all the way for me - but I'm experienced enough to use it, and would be greatly disappointed if this weren't an option for the more advanced users. Algorithms have benefit to those without the requisite theory to understand why certain fingerings work for certain passages, but they do have their limitations. The more advanced you are, the more frustrating and unworkable the quirks of the algorithm become.
Nicholas
Posts: 13137

Post by Nicholas »

Electrode wrote:Given my experience with Midi Illustrator, I would never use an automatic fingering generator on its own...
I suppose "on its own" is fair, but I wanted to dissuade you from swearing off algorithms altogether. While we haven't seen Frost in a while, there has been a lot of work over in this topic (unfortunately, the attachment in the top post isn't the most recent version).

While it's certainly not the most difficult piece (and taking just the (monophonic) right hand certainly makes things easier), around page 3 they ran his program against the first Hanon exercise. The entire piece at once. It mirrored the originally published fingerings 227 notes out of 228.

It could just be possible that MidiIllustrator's algorithm isn't very good. :D
Electrode
Posts: 195

Post by Electrode »

I have seen Frost's algorithm, and I am impressed with what has been achieved, and certainly with the results you cite. However, certain things need to be changed about it before I would use it regularly. I have used it out of curiosity to finger certain passages of a piece by Bach, and I was pleasantly surprised with the results. Still, it's nothing I couldn't have come up with by myself (although I may be a rare case, being an advanced student using Synthesia). Obviously, monophonic right hand fingerings only (for the time being) is a big disadvantage to me given my experience, and the spellings of the notes (counting from A instead of C) strikes me as bizarre. It certainly flies in the face of accepted pitch naming conventions.

However, seeing it fully integrated into Synthesia would definitely be a huge coup for the program, once we get closer to a state of completion with it. Of that, I have no doubt at all. Bring it on! :D
vicentefer31
Posts: 899

Post by vicentefer31 »

cairnz wrote:HAha awesome. <- in reply to the gps image.
Thanks cairnz.

I should make a post with my best pics-collage :lol:
Picasso: I am always doing that which I cannot do, in order that I may learn how to do it.
Kasper
Posts: 149

Post by Kasper »

My main assertion is still that unless you're manually entering from some other piece of reference material (e.g., a book or something), the person that wants the fingerings the most is in the weakest position to provide them by themselves. So to me, manual entry is of fairly limited value.
What about all the teachers that are going to use synthesia for their pupils. For them it would be easy.
English was my worst subject on school, so my language could be a bit awkward sometimes...
Electrode
Posts: 195

Post by Electrode »

I agree. I'm planning on using Synthesia with a six-year-old student. She hasn't responded well to even simplified traditional teaching methods, has a propensity to only want to use one finger on each hand to play anything - and needs a lot of coaxing to even consider trying to play with her left hand, even though we'd only be doing simple things containing one or two notes in the right hand, and one note in the left hand, per bar - hands separately and then together. She wants to eventually do grades, so we're going to have to work for a while to get up to it. (She's very driven, but can also be a little stubborn sometimes!) Nevertheless, she has some musical ability already, because she's picked up little melodies like "Mary Had a Little Lamb" and the first half of "Au Clair de la Lune" without any trouble (right hand only for now).

She is being taught out of a book specifically written for children. It uses sheet music exclusively, with no note names written in on the music. (I'm quite surprised that people actually think this stuff works on children. Teaching methods haven't changed even slightly since I was taught as a kid? We're still chucking the poor children in the deep end and telling them to sink or swim? That's not right.)

My student knows what the finger numbers are, and she knows what the note names are within a five-finger range in both hands (starting on C) - but she can't recognise notes on the sheet music yet without the letters being written in. (I've been writing in the notes for her above or below the staff, in an effort to encourage learning by association.) I'm thinking Synthesia will be a better help to her than trying to understand things she is not ready to understand yet. At least using Synthesia, she will be able to play simple things using Practice Mode, and therefore get a lot more satisfaction out of her lessons.

I want to concentrate on just playing tunes and getting her used to the keyboard for now, rather than trying to teach her too much at once (which traditional methods still seem to want to do). I'm thinking I'll leave the theory alone until she's a little older. I'll re-evaluate her position regularly.

I've never taught a six-year-old before, so this is a completely new experience for me, but if Synthesia pulls through, it's going to become a regular addition to my teaching technique. To that end, fingering would definitely help - with children this young, they're not going to remember everything I tell them, especially since they're not prone to want to practice at home at this age. They'd be better off following a "play what you see" (i.e. immediate feedback) method, rather than the traditional "decode the music and then play it" (which is nowhere near immediate at such a young age as sight-reading ability needs time to build up). My hope is that Synthesia cuts out the "memorising instructions and decoding sheet music" parts, and concentrates totally on fun, playing songs she knows, and getting some real enjoyment and motivation to try and remember what she learned when she gets home so she can show off to the parents!

Even though I can most likely get by without any sort of fingering feature in the program at the moment, it would be nice to see eventually as I anticipate using Synthesia a lot more regularly, and not just for myself.
Nicholas
Posts: 13137

Post by Nicholas »

Electrode wrote:Even though I can most likely get by without any sort of fingering feature in the program at the moment, it would be nice to see eventually as I anticipate using Synthesia a lot more regularly, and not just for myself.
Outside of this Rock Band push, fingering (much like vicentefer's picture depicts) is my highest priority. TonE came up with a reasonable way to store manual fingerings in the metadata. I'm less leery of doing it too, as I get further along with Synthesia translating MIDI into a useful, malleable format (for example, the recent distribute-channel work dramatically changes how I'd have to map those fingerings back to the notes... track-splitting will introduce another major change). Beyond that, I'd love to have automatic fingering that covers maybe 95% of cases. (I'm not greedy. :D )

Otherwise, keep us informed how well-received Synthesia is with your student. The couple of kids I've been able to get in front of a piano with Synthesia make it look like it's especially accessible... though, compared to sheet music, pretty much anything would be.
User avatar
cairnz
Posts: 182

Post by cairnz »

I am thinking of getting a lighted piano for my daughter in a year or so. I think she'll respond well to the lights. As she's only 18 months now, it's mostly just hammering on the piano, but as long as she thinks it's fun, i'm all for it. She sometimes try to imitate what i do (for example press with one finger) and she understands how to do it quiet and loud, it's kinda cool :)
Nicholas
Posts: 13137

Post by Nicholas »

Lighted keys are more fun than I expected them to be. In the long run they're not especially viable (there must necessarily be a full human-response-time cycle between every key press), but for just getting started, I was pleasantly surprised. If Synthesia peels away one of the major complexity layers toward getting started with music, lighted keys go one step further than that. At that point, you're basically just solving the IK problem and getting an awesome musical feedback reward each time. That has to be great for very young children. If only to teach hand-eye coordination and nothing else, it's still pretty awesome.
vicentefer31
Posts: 899

Post by vicentefer31 »

Yes, Lighted keys looks fun even for cats :lol:
and if Nicholas can join the lighted keys with the Free Mode/Donuts Mode, I think it's going to be very funny for very little children like cairnz' daughther.

Edit:
Nicholas, I know now why Shynthesia doesn't have rats ;)
Picasso: I am always doing that which I cannot do, in order that I may learn how to do it.
Nicholas
Posts: 13137

Post by Nicholas »

I'm totally going to teach that cat how to play the piano! Keyboard Cat will have nothing on him!
vicentefer31
Posts: 899

Post by vicentefer31 »

In a couple of month your cat can win American Idol, no doubt!
Picasso: I am always doing that which I cannot do, in order that I may learn how to do it.
Locked