Freesound Forums

What I mean is a sound to be uploaded to freesound . . . Is there a minimum length of time required (must be more than 5 miilseconds or something like that?)
just curious

If you have an interesting "glitch" to post, you can pad it with a second of silence. Anybody who may want to use it should be able to edit the silence off if needed.

À sound cann't be to short.

Thanks for the replies everyone!
@strangly_gnarled, That's a a good suggestion and one I hadn't considered.
I'm used to working/thinking about sound (by that I mean it's a little hobby I have) in the sense of loops and recordings, which you make neat and tidy. I can see why you might need to 'buffer' a sound with silence ....I've just never had the concern that a sound might be too short to be processed (by software that is, not the human ear or brain )

@Klankbeeld That was my real concern thank you

@toiletrolltube Thanks for the extra research there! confirming what Klankbeeld said, a sound can't be too short (for freesound) and it would have been impossible for me to create a sound too short in audacity. (A sample is a measurement but I didn't know what it meant until now so I just avoided it and went with miliseconds)
As for ideas, I just asked a question and got more inspired so thanks yerself

Anyone got the patience to try?

If somebody uploaded 65536 single-sample files, 0000000000000000 to 1111111111111111, (preferably as a pack), it would be the ultimate resource because it could be assembled into absolutely any sound at all! - in 16bit resolution anyway. (I think going for 48bit is a bit ambitious first time around)

The only problem - it might raise copyright eyebrows if you just happened to coincidentally end up with something identical to, say, "Sergeant Pepper" because your computer modelling software for sequencing samples based the "monkeys-with-typewriters" theory proved a little too successful!

My maths is a touch rusty - now I've grown old, but if ever I get around to working out the odds I promise to update here.

Wibby

@strangely_gnarled: Only 65536! Why hasn't anyone done this yet!?!

I am intrigued by the modelling/sequencing software you hint at . . .
The first time I heard of audio/music being put together with samples and algorithms was this track on soundcloud. It's entirely made from samples from freesound and the description about how it was done is really interesting.

https://soundcloud.com/evanxmerz/glass-manufactory

luckylittleraven wrote:
@strangely_gnarled: Only 65536! Why hasn't anyone done this yet!?!

I am intrigued by the modelling/sequencing software you hint at . . .
The first time I heard of audio/music being put together with samples and algorithms was this track on soundcloud. It's entirely made from samples from freesound and the description about how it was done is really interesting.

https://soundcloud.com/evanxmerz/glass-manufactory

I think what strangelygnarled is saying is: create a sample 65536 samples in lenght that contains all the different possible digital values for a 16-bit waveform.
Sounds more difficult than it actually is. You can do it by follwoing the steps below:
At a typical 48,000 samples per second sampleing rate, 65536 samples lasts 1.36533s.
I am sure you are familiar with the Tone Generator function in Audacity. Go there and generate e tone with sawtooth wave form, amplitude =1, Frequency 0.73242 (i.e., 1/1.36533).
Your tone must be at least 1.36533s long, but I recommend instead making it a bit longer, then trimming the excess.
So, once you generated your tone, zoom in at the end portion until you see individual samples (they show up as little dots) and cut it at the end of the first wave cycle.
There you have it - A "sound" that contains every single possible sample value @ 16 bits.
Won't sound as much, since you cannot hear a sound below 20Hz. If you play it all you will hear is a loud pop at the end of the sawtooth wavecycle (which is actually not heathy to your speakers, by the way)

Now, in theory... you could pick samples from this "sound" and create any sound that is possible at 16-bits. But that is an academic discussion.

There are many forms of so-called algorithmic composition.
And plenty of programs around that do that, some of which are free.
Perhaps the simplest ones are those that create sequences of notes (which can be exported as midi files) - These sequences are created from different types of mathematical algorithms.
Other types of programs do things more similar to the example you posted - they create sequences from a set of predefined samples.

I have made some experiments in this area. Usually by constructing very complex feedback loops (sometimes the loops handle sound signals, other times they can even handle midi signals).
If a simple sequence of notes is fed into these virtual circuits (sometimes a single note) they can go on forever (or for a longtime) generating sound.
Some of the sounds are musical or quasi-musical. Others are... erm... I don't know what they are...

HaaHaHaa...

AlienXXX has illustrated the principals perfectly.

Well my suggestion was actually a bit tongue-in-cheek, being fully aware of the impracticality of stacking individual bits(well Bytes actually,)together to construct every possible sound, of any length, that has ever been made - or not made!

Very happy to have stimulated some thoughts though.

Don't know much about cryptology or data compression, but I'm sure I've read somewhere that any numerical sequence, repeating or open (infinite) can be totally characterised (described) by it's own unique mathematical equation. Sine, square and sawtooth are quite simple, and the equation requires less space (bytes) than a full cycle sample of the waveform itself, but the equation for Sergeant Pepper in 16bit 48k resolution would likely require more factors than the number of photons that have hit the earth in the last 4 billion years. Cloud computing still has a way to go.

Got to go and find my brain brush now. Untangle a few trillion knotted synapses!

Wibby

@AlienXXX nice detailed info, I did try that in audacity.... a sample is really small! Any particular reason for such a low frequency?
After reading up a bit on algorithmic stuff I have decided it's fascinating, but a bit beyond me at the moment. I have discovered enough new freeware and info to keep me going for a while.... my brain hurts a bit now.