Great video! Although I got the idea of mirroring, I don't really understand why there are many vertically mirrored spectral elements? It seems like only those above the Nyquist freq should me mirrored, while at 17:07 and 20:36 the components much more below the Nyquist frequency are also mirrored?
The reason being is that all sounds that are not sine waves have overtones and it is the aliased reflections of those overtones that you see on the spectrogram. I am now going to explain it more precisely, but this does mean that I'll first have to explain some basics, namely overtones and Fourier analysis. When you compare the sound of a square wave or that of a clarinet to that of a sine wave at the same frequency, you will notice that they sound noticably different. However, it is just a wave and they have the same frequency, so the conclusion is that in order for these waves to sound different, they must have a different wave shape and this is the case not just for square and sawtooth waves whose name makes it obvious, but also for all instruments. A piano creates a different wave shape than a trumpet which creates a different shape than a guitar and so on. But why exactly are these shapes different? This is where Fourier and overtones come in. Fourier analysis says that every wave shape, no matter how complex, is just a sum of sine waves of different frequencies. And that is the way instruments (or birds) create their specific sound: The vibration, be it a string or air, happens not just a the fundamental frequency such as 110 Hz (which we would typically refer to as the note that is being played), but also at multiples of that frequency such as 220, 330, 440, 550 and 660 Hz and so on, even beyond the Nyquist frequency. This is referred to as the overtone series or harmonic series. Each instrument has its own "signature" harmonic series, so maybe the 5th harmonic is very loud while the 2nd harmonic is very quiet, and this is what creates that instrument's sound. The same goes for birds as well, and their sound is made up of the fundamental frequency and many overtones, and the overtones that are above the Nyquist frequency get mirrored downwards, causing what you perceived as a mirroring of the components below the Nyquist frequency. ETA: These mirrorings then act like overtones, causing the fuzzy sound. Btw, the fuzzy sound of electric guitar distortion is also just created by electronic circuits adding additional overtones.
@@russcharif4320 You actually can't for sure tell if there is aliasing present. Aliasing means information was lost. This might sound like a nit pick, but it's actually true. This is exactly what makes aliasing so troublesome. That said, one way to recognize aliasing in a resampled recording is to look for zig zags in the spectrum. Again, those could have been there in the original sound, really no way to 100% be sure, but it should raise your suspicion.
Video is way too long and not really good at explaining anything tbh. I just wanna know what it sounds like, period, so I can see if my recordings have it. Waste of time.
![If you can't hear this then you're not an audiophile [See description for link to followup video]](http://i.ytimg.com/vi/48PxwspDvrE/mqdefault.jpg)
Great video!
Although I got the idea of mirroring, I don't really understand why there are many vertically mirrored spectral elements? It seems like only those above the Nyquist freq should me mirrored, while at 17:07 and 20:36 the components much more below the Nyquist frequency are also mirrored?
The reason being is that all sounds that are not sine waves have overtones and it is the aliased reflections of those overtones that you see on the spectrogram.
I am now going to explain it more precisely, but this does mean that I'll first have to explain some basics, namely overtones and Fourier analysis. When you compare the sound of a square wave or that of a clarinet to that of a sine wave at the same frequency, you will notice that they sound noticably different. However, it is just a wave and they have the same frequency, so the conclusion is that in order for these waves to sound different, they must have a different wave shape and this is the case not just for square and sawtooth waves whose name makes it obvious, but also for all instruments. A piano creates a different wave shape than a trumpet which creates a different shape than a guitar and so on.
But why exactly are these shapes different? This is where Fourier and overtones come in. Fourier analysis says that every wave shape, no matter how complex, is just a sum of sine waves of different frequencies. And that is the way instruments (or birds) create their specific sound: The vibration, be it a string or air, happens not just a the fundamental frequency such as 110 Hz (which we would typically refer to as the note that is being played), but also at multiples of that frequency such as 220, 330, 440, 550 and 660 Hz and so on, even beyond the Nyquist frequency. This is referred to as the overtone series or harmonic series. Each instrument has its own "signature" harmonic series, so maybe the 5th harmonic is very loud while the 2nd harmonic is very quiet, and this is what creates that instrument's sound. The same goes for birds as well, and their sound is made up of the fundamental frequency and many overtones, and the overtones that are above the Nyquist frequency get mirrored downwards, causing what you perceived as a mirroring of the components below the Nyquist frequency.
ETA: These mirrorings then act like overtones, causing the fuzzy sound. Btw, the fuzzy sound of electric guitar distortion is also just created by electronic circuits adding additional overtones.
@@charlotteice5704 Got it, thanks a ton
Great video👏
CAN WE SEE ALIASING ON MY RECORDING ON SPETROGRAM ON ADOBE AUDITION ,,? OR WE NEED SPECIAL SOFWARE ,,,?
Yes, any software that displays spectrograms would show the signs of aliasing described in the video, if a recording is aliased.
@@russcharif4320 You actually can't for sure tell if there is aliasing present. Aliasing means information was lost. This might sound like a nit pick, but it's actually true. This is exactly what makes aliasing so troublesome.
That said, one way to recognize aliasing in a resampled recording is to look for zig zags in the spectrum. Again, those could have been there in the original sound, really no way to 100% be sure, but it should raise your suspicion.
Video is way too long and not really good at explaining anything tbh. I just wanna know what it sounds like, period, so I can see if my recordings have it. Waste of time.
You're an 1diot.
You're an 1di0T