Relationship of sound timbres and waveforms, is there no difference?
-
- KVRist
- Topic Starter
- 406 posts since 21 Mar, 2015
I'm not sure if something similar was asked before but I wanted to ask about this. Does a sound have unique properties other than its output waveform?
For example, If I made a recording of something like a sea wave, and I took one single cycle waveform of that recording. Would it create the exact same sound if the waveform shape was drawn in a synth (not including the actual waveform sample)?
I know that sytrus can import single cycle waveforms into its operators, but what if the operator was drawn to be the exact waveshape (in theory) without importing the waveform, would it sound the exact same as if the waveform was imported?
btw I am only talking about the single cycle waveform timbre
For example, If I made a recording of something like a sea wave, and I took one single cycle waveform of that recording. Would it create the exact same sound if the waveform shape was drawn in a synth (not including the actual waveform sample)?
I know that sytrus can import single cycle waveforms into its operators, but what if the operator was drawn to be the exact waveshape (in theory) without importing the waveform, would it sound the exact same as if the waveform was imported?
btw I am only talking about the single cycle waveform timbre
- Beware the Quoth
- 33175 posts since 4 Sep, 2001 from R'lyeh Oceanic Amusement Park and Funfair
FWIW, something with as noticeable a proportion of noise as ocean wave sounds, Im not sure if a repeated single cycle is going to properly reflect the randomness of the timbre.phace wrote: For example, If I made a recording of something like a sea wave, and I took one single cycle waveform of that recording.
my other modular synth is a bugbrand
-
- KVRAF
- 2279 posts since 20 Dec, 2002 from The Benighted States of Trumpistan
If you drew at the same resolution the wave was recorded in and drew precisely the same shape, you'd reproduce the waveform exactly, and there would be no difference. The imported waveform is just raster data, and you'd have reproduced it as vector data; the real difference is that vector data can take up a lot more space to exactly reproduce the same raster data, so there would be no point in doing so. It's much better to reproduce something closely so that you can ignore the bumpy bits that break the ideal curve -- much smoother results, much lower resources to store.phace wrote:I'm not sure if something similar was asked before but I wanted to ask about this. Does a sound have unique properties other than its output waveform?
For example, If I made a recording of something like a sea wave, and I took one single cycle waveform of that recording. Would it create the exact same sound if the waveform shape was drawn in a synth (not including the actual waveform sample)?
I know that sytrus can import single cycle waveforms into its operators, but what if the operator was drawn to be the exact waveshape (in theory) without importing the waveform, would it sound the exact same as if the waveform was imported?
btw I am only talking about the single cycle waveform timbre
I speak from experience; vectorizing bitmaps is something we font guys do more we want, but it's a great way -- sometimes the only way -- to revive beloved works from the past. You have to decide which bits are relevant and which bits are due to metal fatigue. It can get really tricky, since the design can be intentionally rough or distressed (e.g. Oldrich Menhart's work), the stamping material often isn't the smooth, strong metal of today (e.g. Albrecht Dürer's woodblocks), the receiving material generally isn't the nice, smooth, fine-grained paper we enjoy today (e.g. papyrus -- not Papyrus, please).
I think you're asking more about how much of a difference there has to be to make a difference. (Actually, you're asking about how close you'd need to be to make any differences meaningless. Same thing, really.) Anyway, I think it's an excellent question. Designers of audio codecs, information theorists, some psychologists and a lot of other people struggle with it daily. It all boils down to how much information you can throw away so that it takes less space and time to store. Me, I rarely hear the difference between a well-encoded mp3 and the original wav, so I don't worry about lossless encodings like flac. (It's amazing how quickly terabytes get used!)
While you're waiting for our responses, why not try it yourself and see what you get? It's take a lot less time than our... let's call it a debate... our debate to come to an answer, and you'll learn a lot of things along the way.
Wait... loot _then_ burn? D'oh!
- KVRAF
- 4821 posts since 25 Jan, 2014 from The End of The World as We Knowit
+1 to all the above.
It may be technically difficult to draw the shape of a waveform for sounds with a lot of noise, because I have noticed that audio editor waveform displays smooth all the detailed noise harmonics beyond a certain resolution (but below the sample rate).
So, even if you had a high-resolution pen tablet to copy the waveform, the smoothed display you copied from would be imperfect. But at ~1kHz, could you notice? And the smoothing would decrease noise, so you might prefer it!
Of course, every waveform in a sea wave recording is different. You could pick a sample, make a wavetable, listen to each of them, and pick your fave. I have done it myself. So, @Jafo's statement above is worth repeating:
"While you're waiting for our responses, why not try it yourself and see what you get? It's take a lot less time than our... let's call it a debate... our debate to come to an answer, and you'll learn a lot of things along the way."
It may be technically difficult to draw the shape of a waveform for sounds with a lot of noise, because I have noticed that audio editor waveform displays smooth all the detailed noise harmonics beyond a certain resolution (but below the sample rate).
So, even if you had a high-resolution pen tablet to copy the waveform, the smoothed display you copied from would be imperfect. But at ~1kHz, could you notice? And the smoothing would decrease noise, so you might prefer it!
Of course, every waveform in a sea wave recording is different. You could pick a sample, make a wavetable, listen to each of them, and pick your fave. I have done it myself. So, @Jafo's statement above is worth repeating:
"While you're waiting for our responses, why not try it yourself and see what you get? It's take a lot less time than our... let's call it a debate... our debate to come to an answer, and you'll learn a lot of things along the way."
H E L P
Y O U R
F L O W
Y O U R
F L O W
-
- KVRist
- 159 posts since 2 Feb, 2017
For most sounds the important characteristics are how the sound changes in shape, frequency, amplitude over time not the waveform at a single instant in time. You can't generally listen to a single wave of any sound but if you could and you had duplicated it's shape exactly that instant of sound would be identical. But it still wouldn't sound a lot like the original which is almost certainly changing all the time.
And in the case of a noise-based sound like crashing seas you'd never find a "single cycle waveform". The sound doesn't really have pitch so it doesn't have cycles.
Steve
And in the case of a noise-based sound like crashing seas you'd never find a "single cycle waveform". The sound doesn't really have pitch so it doesn't have cycles.
Steve
- KVRAF
- 4821 posts since 25 Jan, 2014 from The End of The World as We Knowit
Sometimes I will extract a single cycle from a noise-based sound sample, pitch it to the root note and layer it under the sample, to emphasise the tone while keeping some of the timbre.
H E L P
Y O U R
F L O W
Y O U R
F L O W
-
- KVRist
- Topic Starter
- 406 posts since 21 Mar, 2015
I understand, of course. It won't sound like a sea wave if the single cycle is only playing, it will sound like some robotic noise. But I was mainly asking about the science, so basically if that robotic noise would sound the same as the other robotic noise (actual waveform sample vs. additive synthesis) if you get what I mean.slipstick wrote:For most sounds the important characteristics are how the sound changes in shape, frequency, amplitude over time not the waveform at a single instant in time. You can't generally listen to a single wave of any sound but if you could and you had duplicated it's shape exactly that instant of sound would be identical. But it still wouldn't sound a lot like the original which is almost certainly changing all the time.
And in the case of a noise-based sound like crashing seas you'd never find a "single cycle waveform". The sound doesn't really have pitch so it doesn't have cycles.
Steve
Last edited by Auplant on Wed Nov 22, 2017 11:55 am, edited 2 times in total.
-
- KVRist
- Topic Starter
- 406 posts since 21 Mar, 2015
this is an enlightening post, thank you I'm going to guess that human ears would find it difficult to hear the difference between rastor and vector data in sound. but I guess noisier frequencies might be more noticable.Jafo wrote:If you drew at the same resolution the wave was recorded in and drew precisely the same shape, you'd reproduce the waveform exactly, and there would be no difference. The imported waveform is just raster data, and you'd have reproduced it as vector data; the real difference is that vector data can take up a lot more space to exactly reproduce the same raster data, so there would be no point in doing so. It's much better to reproduce something closely so that you can ignore the bumpy bits that break the ideal curve -- much smoother results, much lower resources to store.phace wrote:I'm not sure if something similar was asked before but I wanted to ask about this. Does a sound have unique properties other than its output waveform?
For example, If I made a recording of something like a sea wave, and I took one single cycle waveform of that recording. Would it create the exact same sound if the waveform shape was drawn in a synth (not including the actual waveform sample)?
I know that sytrus can import single cycle waveforms into its operators, but what if the operator was drawn to be the exact waveshape (in theory) without importing the waveform, would it sound the exact same as if the waveform was imported?
btw I am only talking about the single cycle waveform timbre
I speak from experience; vectorizing bitmaps is something we font guys do more we want, but it's a great way -- sometimes the only way -- to revive beloved works from the past. You have to decide which bits are relevant and which bits are due to metal fatigue. It can get really tricky, since the design can be intentionally rough or distressed (e.g. Oldrich Menhart's work), the stamping material often isn't the smooth, strong metal of today (e.g. Albrecht Dürer's woodblocks), the receiving material generally isn't the nice, smooth, fine-grained paper we enjoy today (e.g. papyrus -- not Papyrus, please).
I think you're asking more about how much of a difference there has to be to make a difference. (Actually, you're asking about how close you'd need to be to make any differences meaningless. Same thing, really.) Anyway, I think it's an excellent question. Designers of audio codecs, information theorists, some psychologists and a lot of other people struggle with it daily. It all boils down to how much information you can throw away so that it takes less space and time to store. Me, I rarely hear the difference between a well-encoded mp3 and the original wav, so I don't worry about lossless encodings like flac. (It's amazing how quickly terabytes get used!)
While you're waiting for our responses, why not try it yourself and see what you get? It's take a lot less time than our... let's call it a debate... our debate to come to an answer, and you'll learn a lot of things along the way.
-
- KVRist
- 236 posts since 30 Oct, 2010
It does not matter how the waveform is arrived at, the same waveform is the same waveform.
However, the sound of the sea is not a single-cycle wave. You would need much longer sample for that. You can still draw it, but to sound like sea you will need to draw a few seconds of it.
So given that there are 44100 samples per second and 16 bit depth means 65536 different values. The odds of arriving at this randomly is 1/(65536^44100). Best of luck!
However, the sound of the sea is not a single-cycle wave. You would need much longer sample for that. You can still draw it, but to sound like sea you will need to draw a few seconds of it.
So given that there are 44100 samples per second and 16 bit depth means 65536 different values. The odds of arriving at this randomly is 1/(65536^44100). Best of luck!
-
- KVRist
- Topic Starter
- 406 posts since 21 Mar, 2015
I know, please see above post. I don't want it to sound like sea, and that was never the intention this was a topic about science of sound, not the actual characteristic of a soundOden wrote:However, the sound of the sea is not a single-cycle wave. You would need much longer sample for that. You can still draw it, but to sound like sea you will need to draw a few seconds of it.