Posted
Could a snippet, optionally, point to a Phrase Pool (or whole Library) so that the Dice for Snippets Memories would have access to the much larger set of source material that can be found there?
Sa., 17.02.2024 - 14:52 Permalink
Under consideration - the Snippets Grid (organized set of pre-chosen phrases), Snippets Memories (which we want to regenerate, or "Dice" at will), and a Library (desired ultimate source of many phrases).
Currently the human must drag phrases from the Library to populate the Snippets Grid. When this is done, the phrases are copied (rather than referenced), and this keeps everything stand-alone and as conceptually simple as possible for both user and developer.
So, how about taking the approach to automate the (slowest) part that the human usually does. IOW, automate the process of populating the Snippets Grid from a Library. Still with copies (not references), but redoable at will.
It is possible to walk thru a Library and identify/select/copy only random "Bass" references, yes? (etc. for each instrument)
UI could be some sort of dialog I suppose, offering for each Instrument in the Snippets Grid a checkbox to be the target of a walk thru of a Library (the Embedded Library (?), any designated Library (?)) which will randomly (or by whatever algo(s) are desired) "drag" phrases for that Instrument into the correct column of the Snippets Grid.
Offhand, this approach seems like it would be doable w/o resorting to indirection. (As a retired programmer, I appreciate the desire to solve a problem w/o indirection, if possible.)
Do you think this might be a plausible approach?
So., 18.02.2024 - 20:32 Permalink
- A library is what it is. Sure, if a library was built from the beginning in 8-phrase chunks then because of Phrase Pool dragging it will be that much easier to use with the Snippets Grid than if it wasn't. Been there, done that - it helps some, certainly. But not all libraries can have this kind of origin. And even for those that do, the gain is only to a certain point and a substantial efficiency deficit remains.
- I'm going to capitalize and bold because I think it's being overlooked - THE MOST IMPORTANT THING OF ALL IS THE USER'S TIME . Anywhere that the user is being asked/required to manually do something 8x instead of 1x (let alone, 16x, or 32x, or 64x !) it is a totally fair, and predictable, ask for the computer to be doing the repetitive job instead of the user. IMO, such requests should not be controversial in the least.
- Remember, issues of the #2 type are multiplied in their painfulness by every user, and every project, and then by all the discussion time/effort around them forever after until and unless fixed.
Granted, there is a list of wishes, and not all can be done at once, but the idea that a request for basic automation of simple repetitive tasks can somehow be considered a defective request - well, that's stunning.
What's the pitch here? "Use Synfire - it's easier/quicker than writing LISP from scratch!" ?
OK, that's a joke.
But if this is a case of "Ugh, I really am not looking forward to writing the code that visits the tree nodes of a Library", then a) I do sympathize, but b) I really think this is a bullet that needs to be bitten! There will be many payoffs from solving it.
(And please forgive me if my technical semi-assumptions are completely off-base!) <g>
Mo., 19.02.2024 - 09:12 Permalink
I understand the appeal of playing a big random jukebox in search for the occasional happy accident. But there are musical and technical challenges with this.
Musical: Using the Snippets grid to audition hundreds of phrases randomly is not ideal. Picking phrases with a chance of 1:500 means you will almost never get the same phrase again, which however is the point when you want to try combinations of phrases (i.e. find out which are working together best).
Without a meaningful preselection you get a random medley with 90% of the results not related to each other and most likely not working.
Technical: There is more to music than drums, bass, chords and melody. Libraries use different sounds than the arrangement. Sounds may play in different registers serving different roles. You could use almost every phrase with every instrument (which is a big plus of Synfire: Phrases are transformed to match pitch ranges).
Therefore it is all but impossible to automate a meaningful mapping of phrases to snippet grid columns. Filling out the suggested dialog or wizard probably takes more time than just grabbing the phrases and dropping them. It just adds another layer of convoluted user interface.
I don't dismiss the idea. I think it has appeal. But there needs to be some form of intelligent preselection, matching or filtering to make random pools of that size useful. There is certainly a way to achieve that but we don't have it yet.
Mo., 19.02.2024 - 14:04 Permalink
As always, thanks for your reply Andre.
Using the Snippets grid to audition hundreds of phrases randomly is not ideal.
What are you suggesting or offering instead?
I try to first frame my suggestions within the existing paradigm of a program to the best of my current understanding.
AFAIAC, using the Snippets grid to audition hundreds of phrases randomly would be an excellent capability to have. I could and would make extensive use of it. I think others would too.
I mean, I could imagine putting a Factory behind every Phrase (no matter where found). If you look at, say, InstaComposer 2, one might say in a certain sense that that is what they've done. Lo, it is good! Is that the approach you are "saving up" to surprise us with? If so, I'm for it ;^)
Without a meaningful preselection ...
That's on me! I've got many good sources of base material for my purposes. In fact, it is a GREAT thing about Synfire that I get to choose and catalog the material that I find worthwhile - as opposed to relying only on a program's hidden algo that I have no access to.
What I need maximal automated assistance with is generating and auditioning combinations.
Re: auditioning, Synfire's Sketches are a superb feature! There is so much going for this program, we just need to identify and remove the roadblocks to highly-efficient workflow. To "complete the pipeline" so to speak.
There is more to music than drums, bass, chords and melody. Libraries use different sounds than the arrangement.
Ah, but this is again on me! If I have a 'Bass' column (Instrument) in the Snippets Grid I only want 'Bass' (Instrument) selections from any library. I'm not aiming to apply a Piccolo Library to my Bass column! (If someone else really wants that, I'll let them take up the torch.)
I'd be very happy with being able to do this explicitly and only with exact Instrument matches. That, as I see it, is "keeping it simple" (as simple as possible, while meeting the goal).
Therefore it is all but impossible to automate a meaningful mapping of phrases to snippet grid columns.
Only for the totally general case you describe. It would be much easier if we are just using the computer to select randomly from "exact matches" (of Instrument).
I don't dismiss the idea. I think it has appeal. But there needs to be some form of intelligent preselection
I am the intelligent preselection! What I choose to "Library" (now it's a verb!), and the Instrument(s) I assign there are the preselection.
--
Filling out the suggested dialog or wizard probably takes more time ...
I beg to differ (though the exact nature of superior UI for this is certainly up-for-grabs).
Here's a guiding phrase: Configure Once, Generate Infinitely
If I can generate (or combine) infinitely, I will never begrudge having to configure once.
P.S. Feel free to use that phrase in your marketing (once you implement it!). Cheers.