05-11-2018, 06:24 PM
had the idea of a "sampler" instrument for a long while even before i started working on the update.
here's my solution to the problems you've stated and some new problems to think about:
for access, i was thinking something along the lines of a new folder in the filesystem. with an extra bit of code and an extra bit of reference to the sequence file, it's possible for anyone to load a sample. basically how the current setup works is that the sequence file references an instrument number along with the volume/delay/reverb/notes/etc. for sampled files, this would be the same, but instead of a simple pointer, it'd be something like "sample:[/app/samples/xyz.ogg]," then the usual stuff. alternatively, it can be another part of data added to the sequencer file itself (which contains things like id/date/owner/etc.)
for the whole sound deal, you're on your own in terms of recording, but the reverb and delay will still be handled by the sequencer like normal instruments are.
for the inappropriate recordings thing, i think sequences would be limited to only one sample and maybe a 2-second recording of at most 128 kilobytes. that's enough for a good recording and there would be no room for more than a small sentence. if someone ends up yelling a curse word and using it as their instrument, the sequence will be deleted, even if the sequence isn't spam. another thing would be potentially loud noises. i think sounds would have to be equalized so that the samples aren't used as an ear-weapon.
here are another set of problems that i'll present to you:
- the sampler instrument is probably going to be .ogg files only, which, to my knowledge, aren't supported on Edge as of yet. this means that samples won't be available to Edge users (if there are any active ones, that is.) i could potentially see if it's possible to use mp3 files exclusively for this feature, but mp3 isn't as compressible (to my knowledge) as is.
- currently, the instrument files work like this: you have a recording with all of the notes being played at equal distance apart at a certain tempo. there's a file called "settings.php" where you tweak some numbers in the file corresponding to the instrument in order to configure its range, volume, tempo (this is the value that calculates where in the audio file it needs to point to in order to play a certain note), and corresponding MIDI instrument.
the first problem is that under the current system, only one note on the player would be available. the recording could be duplicated many times over and its pitch raised/lowered, but that presents the problem of selecting a base note or having one already added. having a base note be automatically added would mean some files might have too high notes or too low notes, so selecting is our route. what we'll most likely do is just have a bit of code that does the whole pitch raising/lowering.
here's an article on how that could theoretically be done:
https://codepen.io/enxaneta/post/frequen...ical-notes
the remaining problem would then be figuring out how to apply that change in pitch to the sample and have it calculated within the client, which could potentially increase loading times for samples.
-finally, the amount of samples that it'll potentially create. since we have many, many active users daily, and many, many sequences. in a month's period (that is, from april-may 11/12,) there were 28,075 sequences. if we let both guests and registered users have access to samples, even if it's only 128k files being uploaded as i proposed, i think that it'll eventually run the server out of space far faster than conventional sequences. even if a tenth of those sequences had 128k files uploaded as samples, you'd still be looking at 359 megabytes of data per month. if it was a third, you'd be looking at almost 2 gigabytes.
i think it's smart to have sampled instruments be a premium feature, but not to the extent that one would have to pay money for it. i'm thinking that the feature can be for registered users only, which would cut down that filesize figure drastically.
these are all of the problems that i can find with having a sampler.
when saving a sequence as a remix, by the way, should the sample remain referenced in the sequence, or should it be unloaded, forcing the user to upload a new sample? to cut down on files, i don't think we'd have it duplicate the file. this isn't necessarily a problem, but a question of a feature. i can see how people would be divided on it if we'd made the decision ourselves before asking you guys about it.
here's my solution to the problems you've stated and some new problems to think about:
for access, i was thinking something along the lines of a new folder in the filesystem. with an extra bit of code and an extra bit of reference to the sequence file, it's possible for anyone to load a sample. basically how the current setup works is that the sequence file references an instrument number along with the volume/delay/reverb/notes/etc. for sampled files, this would be the same, but instead of a simple pointer, it'd be something like "sample:[/app/samples/xyz.ogg]," then the usual stuff. alternatively, it can be another part of data added to the sequencer file itself (which contains things like id/date/owner/etc.)
for the whole sound deal, you're on your own in terms of recording, but the reverb and delay will still be handled by the sequencer like normal instruments are.
for the inappropriate recordings thing, i think sequences would be limited to only one sample and maybe a 2-second recording of at most 128 kilobytes. that's enough for a good recording and there would be no room for more than a small sentence. if someone ends up yelling a curse word and using it as their instrument, the sequence will be deleted, even if the sequence isn't spam. another thing would be potentially loud noises. i think sounds would have to be equalized so that the samples aren't used as an ear-weapon.
here are another set of problems that i'll present to you:
- the sampler instrument is probably going to be .ogg files only, which, to my knowledge, aren't supported on Edge as of yet. this means that samples won't be available to Edge users (if there are any active ones, that is.) i could potentially see if it's possible to use mp3 files exclusively for this feature, but mp3 isn't as compressible (to my knowledge) as is.
- currently, the instrument files work like this: you have a recording with all of the notes being played at equal distance apart at a certain tempo. there's a file called "settings.php" where you tweak some numbers in the file corresponding to the instrument in order to configure its range, volume, tempo (this is the value that calculates where in the audio file it needs to point to in order to play a certain note), and corresponding MIDI instrument.
the first problem is that under the current system, only one note on the player would be available. the recording could be duplicated many times over and its pitch raised/lowered, but that presents the problem of selecting a base note or having one already added. having a base note be automatically added would mean some files might have too high notes or too low notes, so selecting is our route. what we'll most likely do is just have a bit of code that does the whole pitch raising/lowering.
here's an article on how that could theoretically be done:
https://codepen.io/enxaneta/post/frequen...ical-notes
the remaining problem would then be figuring out how to apply that change in pitch to the sample and have it calculated within the client, which could potentially increase loading times for samples.
-finally, the amount of samples that it'll potentially create. since we have many, many active users daily, and many, many sequences. in a month's period (that is, from april-may 11/12,) there were 28,075 sequences. if we let both guests and registered users have access to samples, even if it's only 128k files being uploaded as i proposed, i think that it'll eventually run the server out of space far faster than conventional sequences. even if a tenth of those sequences had 128k files uploaded as samples, you'd still be looking at 359 megabytes of data per month. if it was a third, you'd be looking at almost 2 gigabytes.
i think it's smart to have sampled instruments be a premium feature, but not to the extent that one would have to pay money for it. i'm thinking that the feature can be for registered users only, which would cut down that filesize figure drastically.
these are all of the problems that i can find with having a sampler.
when saving a sequence as a remix, by the way, should the sample remain referenced in the sequence, or should it be unloaded, forcing the user to upload a new sample? to cut down on files, i don't think we'd have it duplicate the file. this isn't necessarily a problem, but a question of a feature. i can see how people would be divided on it if we'd made the decision ourselves before asking you guys about it.
-
x1
-
-
x1
- Firebolt391d
-
