Why Do Instruments Sound Different? The Science of Timbre
Instruments and music have been around for years, with many instrument makers knowing the art to perfecting the sound coming out of an instrument. However, in this video, I'd like to show the science as to why instruments sound different, and what we can learn from it.
3Blue1Brown video:
• But what is the Fourie...
Blind violin test:
www.sciencemag.org/news/2017/...
The characteristic sound of an instrument is known as timbre, so violins sounds different to trumpets because they have different timbres. This is something that seems obvious, but is used so much in orchestral music, where certain timbres can evoke certain emotions or qualities. In terms of a sound wave, timbre can be thought of as the overall shape of a wave, with different instruments creating different waves. My original question as to why timbre should exist still stands, however, as why should a vibrating string create a complex pattern like this, when we are taught that vibrating strings look smooth and sinusoidal, and why should a vibrating string create a different pattern to a vibrating column of air? And also, why do expensive or old violins typically sound better, what is it about these waves that improves the quality of a sound? These are all important questions with more physics behind them than you might realise, and my goal in this video is to at least try to convey how ordinary processes in our lives can be broken down into deep and fundamental questions concerning physics. Let’s start by taking a closer look at these waves.
When we listen to a note coming from an instrument, the sound that we hear is not just the note pitch which we recognise, but also consists of many, much quieter harmonics. For instance, a violin playing an a natural will not only be producing that a natural note, but also other harmonics with different pitches. These harmonics add together, in a process known as superposition, to form this messy wave which we can see now. The only problem is, it’s hard to tell what harmonics are in the sound by just looking at the wave, and this is something we will need to do to interpret why instruments sound different. Luckily, there is some neat maths which can help to turn transform a messy signal into the component harmonics, a tool known as the Fourier transform. Fourier analysis is a topic which can take up an entire video, and 3blue1brown has a fantastic video which explains it without any maths, I’d highly recommend watching that if you’re interested. The fundamentals of Fourier transforms are that they take a signal in the time domain, so a graph with time on the x axis, and transform it to a graph in the frequency domain. What this means in terms of waves is that you can take a messy signal like this one, and apply a Fourier transform which will unravel this signal, and determine what frequencies of waves were added together. This corresponds to a graph with frequency on the x axis, showing how much of each frequency of wave was added to make the original signal. Using this, we can determine the intensity of each harmonic by translating frequency into pitch, and so gain an idea as to what harmonics are in any sound, and how loud they are. For example:
The Fourier transform is an incredibly powerful technique which is used in many areas of physics, and it’s in applications like these where it proves just how useful it is.
Using this, I was able to find samples and analyse them. The following clips show the instruments and their respective Fourier transforms.
why do instruments sound different
what are harmonics in physics
fourier transform in music
violin fourier transform
why do violins cost so much
why do stradivarius violins sound better
stradivarius violin vs normal
science of sound waves
science of music and sound
how do instruments work
how do instruments make sound
how do instruments sound different
helmholtz string,chladni plate
chladni plate music
chladni figures
how do chladni plates work
music science
Пікірлер: 47
I've been looking through a lot of videos on timbre and this was one of the most informative. This was really good, thank you. Hope you're able to keep making videos!
Duuuuude dude!!! you are amazing. great contant. You are gonna explode soon :) proud to be here before
This was very helpful, please make more videos about the science of music.
this is so good! can't believe you only have 1500 views. i loved it.
wow, this is something ive thought about on and off for years and finally decided to look it up. this video is amazingly well done. it covered all my questions and more, and was explained in such a simple way. thanks for putting the time in for us.
I loved it so much. Thank you so much! This really helped me understand and.. idk how to describe it, I hope your style doesn’t change. The standard style leaves me confused, or a little bored, but it seems like maybe you just did what you yourself like and, it made a really good experience. It had me interested, feeling a warm and fun atmosphere, not confused, not bored, it was easy to follow and think through it. Thank you and please don’t change as you get more experience and try to shape yourself up! It is so beautiful this way
superb explanation! this is an excellent presentation
This video should have more views! great content!
Excellent, just what I was looking for.
Great Information Thank You...
Superb video and explanation! Thanks!
thank you very much, I teach this and video is full of information and well made
Loved this video!
the correlation between the patterns on the vibrating metal plates and the Schrödinger's equation is SOOO mind blowing, fascinating, magical and just overall incredible, leaves me speechless in front of the beaty of nature, the connection with the universe within itself at a micro and macroscopic level🤯great work mate!
really intresting nice video! any knowing obout "how diffrent materials composition effect on the timber"?
great video.. thanks for this
thank you very much this is very informative
Thank you very much. This was really good.
Why in orchestras they use more string instruments than wind instruments?
@robertboyle4768
Жыл бұрын
It's an issue of the overall balance of the volume each section can provide. Pretty much any wind instrument can play louder than any string instrument. So in order to have the musicians play at a natural volume for their instruments, an orchestra is just going to need larger string sections than wind sections to create that balance.
Very useful! I’m am writing about tone production and would like to be able to measure wich overtones are more or less present when I use the bow in different ways. Will look for the «Fourier transform»!
Well done friend.
Nice explanation, but a bit dumbed down. In the case of the violin, it's not just the string that causes air to vibrate but also the body of the violin, and e.g. the bridge/comb that holds the strings up (at about 06:00 the different resonant patterns of two violin tops, giving different timbre, are explained a bit). Sections in the curved top vibrate and the violin maker uses toy like mini planers to shave the curves but also adjust resonant properties. Before that (a)he actually acoustically tests the boards that the instrument is made from, for musical suitability given the desired result. The top may have a stud coupling it to the back and with the stud as frequency filter, this adds different "harmonics". This so far is based on "contact" between vibrating elements where one passes vibrations on to the other. There still is the "through the air" passing on, that e.g. we hear a lot in pianos - not just with the sustain pedal down, because the top two octaves have no dampers. Second, timbre is an aspect of the human voice as well and explains why you can hear the distinction between e.g. father and mother's voice. Third, the video's very busy wave patterns don't really help in understanding timbre - better explained with the simple graphs that show a modified sine or block wave as a thin line depicting pressure as function of time. Fourth, the human ear is not a single sensor device, but has some 20,000 frequency (band) tuned sensors that measure at different frequencies. While the ear's anatomy and geometry in the case of an individual may define their audible frequency range, these ~20,000 sensors allow us to hear timbre instantly. In electronics terms, the bandwidth needed for a 440 Hz (piano's central A in today's tuning) sine wave is about 440 Hz, but if we faithfully want to reproduce the piano's timbre, at that 440 Hz, then we need - expressed in sine wave frequency - orders of magnitude more. Being able to play back 20,000 Hz sine wave is simple. Playing something that looks in the oscilloscope like a block wave in a way that precisely sounds like a block wave, even at 440 Hz is darned difficult. While this is an easy problem in Fourier, it is extremely hard to design/build devices that can do this and I'm not even considering, say, a 3-way loudspeaker that cannot reproduce timbre properly towards a random listening position (in the room) - so many cans of worms and rabbit holes. Inertia and momentum of molecules in gas and of electrons in a copper cable play a big role here, and even quantum things around the cables where most of the signal travels play a role there. By the way, in humans, hearing is much faster than seeing and as long as the slow travelling of sound is not an issue, it really works out that way, versus light speed. We have comparator networks in the brain to analyse left/right differences and as some neurons have electrical synapses rather than enzyme-modulated biochemical ones, these have lower latency and we can imagine these electrical neurons in the audio paths of hearing. Fifth, "hearing" and recognition must be learned and take significant time for us to become predictably reliable at that. Therefore we need years before we can recognise voices (characteristic to individuals timbre) and therefore some of us hear a "piano" when a pianist test-plays 5 Steinway D individuals, and coming back next year is able to say, I played this one last year.
Oohhhh …..I understood the maths of Fourier series 50 years ago . Now I want to improve my violin playing and why my small collection of violins sound so differently. The work of Carleen Hutchins is used today for plate tuning of violins and takes the mystique out of violin making.
Very cool video--the KZread algorithm led me right! Thank you for the excellent content. :)
It would be informative to start with a pure tone and add harmonics sequentially to build up a complex timbre. Also, the intensity of the dominant frequencies (peaks) in your diagrams is shown on a logarithmic scale and as a result it is difficult to discern that the intensities actually vary quite a bit. Maybe by adding frequencies to build up a complete musical note you could inform the viewer of the subtlety of the added subordinate frequencies. Also, what about all the noise between the peaks? What does that add to timbre? Thanks for posting this. I looked this up because of a recent article in the science journal "Animal Cognition" that determined that dogs could perceive both pitch and timbre in a human voice, and I didn't know what "timbre" was.
@drdca8263
2 жыл бұрын
I agree, that would be interesting, and I’d like to see a video showing how specifically the different harmonics of different instruments change the sound, as well as whether the noise in-between matters. I suppose the way to see this would be to take a spectrogram , or just one of these Fourier transforms, and modifying it, and then taking the reverse operation. Iirc the different vowel sounds are determined by the differences(ratios? Idr) in the frequencies of the first few peaks. In contrast, I think with instruments, I think (but am not sure) that the difference in frequencies of peaks is generally basically the same between instruments, and that it is the relative volume(height) of the different peaks which is different between different instruments? So like, a pure sine wave should have a single peak, with the other peaks having a height of zero, but a triangle wave or square wave should have peaks in the same places as each-other, but with different patterns in the heights. (Though I think it should also be possible to express e.g. a sine wave or square wave in terms of triangle waves, and this would just be a change of basis. Hmm..., So, the space of periodic bounded signals, should be L^{\infty}([0,2 pi]) I think, and the Fourier transform should send that to... what, \ell^1(\bz) ? Or... Uh, L^2([0,2pi])is much more convenient to take Fourier transforms of, because I’m pretty sure it should give l^2(\bz) , and these are both Hilbert spaces ... But I guess anything bounded will be in L^2 as well, So, uh, will this change of basis from sine waves to triangle waves be unitary? I guess the question is, what is the inner product of triangle waves with different frequencies, where one frequency is a rational multiple of the other? Hm, I’m guessing it generally isn’t zero? Or, maybe it is? Hm. I should get some scratch paper and work this out.)
Such an underrated and amazing video. One question: once transforming to frequency domain, does it imply that a new plot is needed for every discrete "time" relating back to the time domain?
@ollienash4124
3 жыл бұрын
Let me know if I’ve got the wrong end of the stick, but if the question is whether you need a frequency plot relating to each specific instance of time on a time plot, then the answer is in general no. The frequency plot will show you the frequencies of all the sine waves making up the original signal, and since the sine function varies with time, you get a intensity vs time plot which varies with time. The frequencies of each of the sine waves are fixed however, meaning that an intensity vs frequency plot does not change with time, assuming that the source is providing a consistent signal, for example a trumpet note being held at the same pitch and loudness, and not changing pitch etc. Hope this helps.
@JaZoN_XD
Жыл бұрын
@@ollienash4124 Don't know why it took me this long to come back to this video and find your response, but yes, that answers it, thank you!!
@californiadreamin8423
Жыл бұрын
@@ollienash4124. Hello . Are you aware of the work of the late Carleen Hutchins ? I am fortunate in having 2 fiddles made using her techniques, and have a photo of the glitter patterns when the top plate was complete. This technique takes the mystique out of violin making, particularly for the amateur maker who only makes a few violins. I’m convinced that Stradivarius had a top plate which he used to determine how close he was to achieving the sound he was looking for , on each fiddle he made. Additionally, the violin bridge can be “tuned” to filter out undesirable frequencies and achieve the sound the maker is looking for. I have read that quite a few bridges are made before the maker….and the player…is happy.
Can we get a solution (number) for timbrs for piano (for EX ) ?
Amazing man just amazing
... excellent ... !!!
I am collecting resource for a personal project where I will be creating a modal that validates a song has a specific instrument on it or not. This video was very useful for me, thank for that. I am planning know more about how to analysis this "Timbre" for the instrument as my next step. Guys please do let me know if you have any other reference that might help, thank you.
@BARATHANM-ln4cd
Жыл бұрын
Thts really great....want to know about the outcome...I too have a similar idea..
Very helpful video. Just wish you had included links for further reading :'|
Nice video, but i can barely hear you sometimes behind the "background music"
Nice
you have fried my brain.
love it
good video
@kadrodindubldva
3 жыл бұрын
I left a like but fix the mic noise
Nice video, but the balance with Claire de Lune was off for me. I feel like it was a little too prominent in parts of the video.
Violin and trumpet….kinda? I’d have gone w/real instruments if putting this much thought into a complex, fascinating topic. Timbral approximations are no substitute. These sound synthesized to me?
that is not right
Bla bla bla, i dont know what timbre is , bla bla bla...