Let's clone an instrument's timbre using nothing but sine waves and mathematics on our first plunge into the deep, powerful world of additive synthesis.
When you said "additive synthesis is the future" I thought "prove it, dude". Later, you played that "cello" clip from one of your tracks. I then thought "well, this dude has certainly proved it". Excellent video, man. Truly fascinating stuff. Thank you.
That's how lossy audio codecs work as well. MP3, AAC, Ogg, etc. store audio data as spectral information and whenever you play any of these formats (e.g., when you are watching a CZcams video), the original sound is recreated (imprecisely) based on the harmonics. That doesn't prove that "additive synthesis is the future" though. The problem is in manipulating the properties of the synthesized sound in a meaningful way.
@@kaktotak8267 that’s where deep learning can help. Retain the signature characteristics of a sound and then manipulate to obtain new sounds with similar characteristics (if i understood correctly). Lossy audio formats do the process in the opposite way, simplifying an existing sound, while here is about generating it without having a source sample but only a machine learning model that predisposes the additive synth to get a specific sound rather than doing that manually as human. This doesn’t mean there is not a musician or sound engineer behind that. It’ll be just the future way to synthesise audio on any type with a standardised tool; other types of synthesis will remain, according to the producer’s taste of course.
"Additive synthesis is the future." I told my friend that years ago after researching the additive synthesis timbre morph capabilities in Kyma and and Alchemy. I don't think any reputable sound designer would doubt that. It's just a matter of getting computers powerful enough to render a FFT fast enough and with enough detail that it eclipses what humans can naturally hear. Just like how our sample rate and bit depth cover the resolution of human hearing. Of course, that bar will be raised once we invent cybernetic cochlear implants that not just restore hearing, but expand it. There's all sorts of fun directions that this discussion can go.
same goes with sight or smell. With cybernetic enhancements, more wavelengths (frequencies) would be visually detectable by humans, now this could either mean new colors, or something completly different that we can't even begin to imagine. And for smell, we could smell certain substances (like gasoline, which is, I'm pretty sure, odorless) instead of adding smelly chemicals to the substance.
Thank you for this video, you've inspired a project that I am pursuing for my Computer Science thesis project. I just got approval from the department today to design an fpga based additive synthesiser. I'll post all about it when I can, but it's about a year from full realization.
As someone without a technical degree or a sound engineering background, just getting into music production, this video honestly shat on my ego and made me want to run in the other direction. Watching it a second time, I really appreciate the complexity, effort, and time that went into making this skillful demonstration and definitely want to start looking into additive synthesis and some of the terminology he covers on my own, so I don’t feel so retarded next time.
The Kawai K5000 was a hardware synth released in 1996. It could do a lot of similar stuff to what is demonstrated here. I have a Cello patch similar in terms of realism, will get around to sharing my patch soon to other K5000 owners.
I hope you end up doing a part 2 to this, as it is a very fascinating subject. Covering how you did the cello and beyond would be fun to listen to (though I don't know if it'd be fun for your to make). Cheers.
This is incredible, basically with the correct maths you can make any sound that you can think of, and you can make incredibly detailed instruments without have massive sample banks, similar to your piano tec video, or is my understanding wrong
Even though synthesizing could require less space in disk since it only require parameters for the sound, the process of synthesizing is heavy for processors, that is why some digital/computer synthesizers usually create a temporary library with all the notes/pitches. Like a pre synthesis of the sound, so it reduce amount required process for rendering and live play. Nice vídeo, thanks for sharing!
Wait, did you _create_ that cello in Starlight? Was that a synthesized cello that you created?? How on earth did you model the bow-on-string attack and stuff?? WHAT AM I LISTENING TO
I think he mentions swapping. So obviously there is the sample and then it crossfades between it and the modelled version, which sounds mangled. Not sure if that's correct, but sounds that way to me.
@@EvilDragon666 I got the impression, both from what he says and the video caption, that the cello is entirely additive. I think he might have used a sample and some sort of machine learning to get it spot on.
@@EvilDragon666 Nope, that's definitely not crossfading. He also mentions he starts randomising the partials that make up 'what you thought was a cello'. Seems pretty clear to me. TBH if the modelled sound was that far off why would he demo it as it'd disprove his entire argument.
likely just adjusting partials with envelopes. the Harmor plugin in FL Studio is an additive synth plugin that i believe lets you do things like this, but it sounds like he's pretty in tune with low-level synth programming stuff so it's hard telling what tools he used for it.
I actually tried resynthesizing a violin in Alchemy, by just importing a random violin piece that was in no way prepared for that, and Alchemy made an additive synth patch that sounded very close to the violin. I believe one could prepare a sample in a certain way for the additive engine to have an easier time analysing and recreating it, so these things are real. I easily believe this was an additive cello.
These videos are so great, man. I'm pretty stupid but still followed. Up until deep learning making sampling obsolete, but that's neither here nor there. I would love to have a video dedicated to that cello sound because my mind is seriously in pieces right now. Actually it'd be cool for you to do little breakdown videos of other things you've used in your music too. Just so we have real world applications in a creative setting- and then an explanation of how you got there.
Hi Ben. Great video. I fucking love your channel. I disagree about "made sampling obsolete" however. I would argue that sampling isn't just about capturing the tone of an instrument. I think this makes instrument sample libraries a thing of the past, but actual sampling, as is the case in a lot of hip hop, for example, is about taking little chunks of a performance, usually recorded with gear you wouldn't have access to, or in a way you personally wouldn't. Sampling can be about creating a musical collage, rather than just digitising an instrument.
I work in Linux, and recently I've gotten obsessed with experimenting with the additive synth "Organic", which is an eight channel additive synth in LMMS. I feel like, like everything else, this is taking me down a road.
Forgive my ignorance but I'm just now realizing I'm watching THEE flashbulb! The lawn wake series got me through a lot of hard times. Thanks not only for the music, but for the education too
Ummm I'm a cellist, and i was like oh great, here comes another weak cello imitati- 11:41 when the pitch bend of a hard cello bowing came through, I was beyond shook I don't know whether to be offended or amazed, I chose fearfully amazed🙏
Dude, thank you for the time you put into this to explain it to us. I get it much more now, the small amount of time I spent with additive synths made me think they were somewhat limited, I was so very wrong!
Not a musician, just a DJ but absolutely in love with your channel. Thank you so much for all your educative work!! It really means a lot to get input from down to earth people who embrace science !!!
DJs mix stuff that they didn't even make at parties, events, they don't program synthesizers to play unique instruments and make songs. The flashbulb is definitely a musician.
@@JiMMy-xd8nu I was refering to myself (the dj) and just praising the musician (benn Jordan) for his amazing educative work. (as a DJ I highly value musicians dedication and passion for music creation).
@@goldfish3385 Hey, my english isn't the best either, and your comment wasn't that confusing, I just stopped functioning while reading it the first time.
Pleaaaaase do more videos about additive synthesis and modelling! The track with the chello is bonkers! I really want to know how to do things like this
Outstanding video, thank you ! I am studying csound in my free time in order to program cool synthesizers. I've got some pretty good results so far, but I hope there will be a video on csound on your channel one day.
I'd like to thank all those mathematicians & engineers whose insight have vested on us a way to understand how systems work. Moreover, all these transforms are all related... Even on quantum mechanics! And Ben, you've done an outstanding job. Thank you.
this was fantastic, thank you im a uni student and have chosen additive synthesis as my main topic for an online presentation. this helped a lot. thanks dude!
Fascinating video, definitely subbed, and at this point in time, I'm seeking it video content like these to really become better at sound design, and understanding how to make anything from basically nothing 🤔👀 thanks for this video tho, learned a lot
I really wish you were correct in being confident about the future, 6 years onwards... I'm still looking for a harmor equivalent that does resynthesis that's not daw locked ;__;
When describing the basics of additive synthesis, it sounds just like using draw bars on a Hammond organ! And your first example of doing a few harmonics sounded just like a B3. But, because the Hammond used physical sine wave generators (tone wheels), it could only be built with just under 100 of them, and their frequencies are tempered, such that the harmonics are shared across every note, blending across all notes played, and especially blended when pulling all the bars out. Then it becomes pretty hard to achieve dissonance and instead there’s a smoothed out harmony of every note played to each other. Great for rock and roll players to just go wild, never a false note, but of course never a true one either (hence cheesiness). I don’t think samplers capture this effect, and I wonder if there are true circuit emulators that do (I don’t follow modern gear enough to know). But your mathematical breakdown of the unique harmonic series frequencies for every note one would need to reproduce things accurately is a nice clarification of how challenging this kind of tech would have been to consider in the time of early electronic and electro-mechanical organs!
Hi, I really like the way you are explaining things in your videos Benn! I am actually trying to create additive synthesis for my master project, so I am really looking forward to seeing the next level tutorial... We can find some almost similar explanations here and there, but one always lack of deeper understanding for going further, in order to reach more realistic results... I am not talking of perfect or high level recreation, but more how to enter in the real world of additive synthesis after the basics... Hope you'll find the time for us man, and long live your channel! Greg
Additive synthesis is also the past of Synthesis. When You made Your dull saw You made a mistake, it was instead a "sharp" saw because You neglected to reduce each successive higher harmonic amplitude by 1 over it's frequency.
I'm not exaggerating, when you said:"combined with deep learning", I jumped around and cheered for 15 seconds. From now on, I'm gonna start recommending this channel to ppl.
What if you had essentially 128 resonators for each key of the piano. Essentially each key is a very sharp EQ correlating to the note. You could also put a gate on each key and a trigger so the amplitude of the individual resonator notes would trigger MIDI on a synth. You could also do things like have each resonator's pitch stretched or arranged in a microtuning so when audio is fed through it, only harmonics relating to the current tuning of this resonator synth can pass through.
When you mention that additive synthesis can be used where disk or memory resources are limited (like games on microprocessors etc.), it sounds like you are reinventing lossy compression (like MPEG/Audio).
I can't understand what you saying but I truly appreciate this video and will be re-watching this to try and grasp each thing that you are saying, because I just started learning and designing synthesis and for some reason I would subtract everything until I got the frequency range that wanted, regardless of dB and just say to my self, I wish I could layer over or bulk up the dB of this range because I have the sounds sculpted to the way I wanted. Kind of a novice, dumb, and borderline unintelligible comment, but the entire video I was saying: I don't understand what you are saying, but to a degree, I get what you are doing, I'm just not formally educated on the matter. Thanks for this! Hopefully I can grasp some of these theories and notions sometime soon so I can explain what I already kind of knew to others! The sine is pretty much the water of sound haha
I love jamiroquai! nice! my third album i ever bought was the album with virtual insanity on it.... i believe it is called Supersonic? been a fan every since and love all their other stuff
Great intro to additive synthesis, very educational (even if i don't completely understand everything)! Loving your content so far! How would you program this in Csound?
Check out LayR synth. It is basically an almost unlimited number of subtractive synthesizers that can be added together making additive synths with more complex waveforms. For instance when you make a bell tone in LayR it is easy to create waves for each overtone. Very cool 👀
As far as I understand is the basic principle of FM Synthesis also to stag layers of sine waves ( especially the feedback operator ) to get such results... Isn't it also a additive synthesis? I've get myself my first FMsynth while waiting for my Tracker and - I admit quite confusing in the begining - but I guess there's a lot potential in it
FM achieves the same result but with different (and I would say more efficient) means. When you modulate one sine wave with a different sine wave as is done in FM synthesis, the resulting waveform is basically a sine wave made of another sine wave. Imagine the shape of a sine wave, then zoom out a little and imagine bending that wave into the shape of a bigger sine wave. The resulting sound is the same fundamental pitch but has more overtones (aka partials) giving it a more vibrant timbre. You could create the exact same timbre by adding the right sine waves together but it would require a lot more sine waves, depending on the ratio used in the FM synthesis. Basically both FM and additive synthesis could theoretically perfectly replicate any complex sound, they do it in different waves. I think of additive as adding sines together and FM as multiplying sines together
Wow! I had no idea of the power of additive synth until this video. I'm now on a mission to emulate a clavinet this way. You said your demonstratrion of capturing the melodica's timbre is "not the most efficient way of doing additive synthesis in 2017." What is a more efficient process?
My thoughts are - additive synthesis is definitely a useful tool, but it is never going to make samples obsolete. Just IMO! You need a ridiculous amount of processing power to handle enough partials (and allow good enough polyphony!) to completely fool us. To emulate a static sound is not that difficult and I mean you can do it in Alchemy (700 partials max). But the big kicker is emulating the actual physics and behavior of a particular instrument (which is easier done with physical modeling than additive synthesis), THAT would take an awful lot of time to get right for each instrument. It took Pianoteq 10 years to get to an acceptable level of quality (and that's not an additive synth - so for additive synth it would be even more difficult/time intensive!). And what about the recording chain? The mic, the room? Samples still portray this most faithfully, and you cannot reproduce this with 1000 partials. Maybe 10000 would suffice. And ultimately: sampling is SIMPLE to use. Additive synthesis is anything but. Yeah you can get to places with machine learning, but that is still not really layman people territory (programming such algorithms), while anyone can grab a mic and a recorder and... do stuff with it. I say all this as a supporter of physical modeling (I absolutely adore Pianoteq), but also with deep knowledge about sampling (since I've been developing Kontakt libraries for the past 13 years).
Wow. Not easy for me to follow, but the content is great and the video is very well done. Thanks. I'd be interested in physical modeling and eurorack patching :)
Fourier series rule 🤓 awesome explanation and demonstration. Not sure if you have used the Make Noise telharmonic, but it is a pretty cool additive synthesizer. It doesn’t have 500 sine waves and can’t do what you did there at the end of the track, but it can recreate lots of sounds well.
hopefully you'll see this, was that final sound achieved in either Alchemy, Csound, or Kyma? they are the only two systems I could understand such a sound being made in... harmor and morphine dont seem like they could do such a smooth "randomisation" of the partials.
Hartmann Neuron synth, we need to bring it to PC! If you haven't heard of it open an new tab and look it up. It's the saddest loss in the synth industry in my opinion. Work is being done to bring it to PC but I think it should be given new life.
That cello to synth morph melted my brain. I'm legitimately intrigued by the possibilities of additive synthesis now.
Same.... I'd really love to see a tutorial on how that process worked
@@neonblack211 exactly, I'd love to hear more about that
@@neonblack211 was there ever a video?
@@ChronicalV no not that i know of
When you said "additive synthesis is the future" I thought "prove it, dude".
Later, you played that "cello" clip from one of your tracks. I then thought "well, this dude has certainly proved it".
Excellent video, man. Truly fascinating stuff. Thank you.
yeah, i almost throw my laptop out of my window when i heard that!
That's how lossy audio codecs work as well. MP3, AAC, Ogg, etc. store audio data as spectral information and whenever you play any of these formats (e.g., when you are watching a CZcams video), the original sound is recreated (imprecisely) based on the harmonics.
That doesn't prove that "additive synthesis is the future" though. The problem is in manipulating the properties of the synthesized sound in a meaningful way.
@@kaktotak8267 that’s where deep learning can help. Retain the signature characteristics of a sound and then manipulate to obtain new sounds with similar characteristics (if i understood correctly). Lossy audio formats do the process in the opposite way, simplifying an existing sound, while here is about generating it without having a source sample but only a machine learning model that predisposes the additive synth to get a specific sound rather than doing that manually as human.
This doesn’t mean there is not a musician or sound engineer behind that.
It’ll be just the future way to synthesise audio on any type with a standardised tool; other types of synthesis will remain, according to the producer’s taste of course.
"Additive synthesis is the future." I told my friend that years ago after researching the additive synthesis timbre morph capabilities in Kyma and and Alchemy. I don't think any reputable sound designer would doubt that. It's just a matter of getting computers powerful enough to render a FFT fast enough and with enough detail that it eclipses what humans can naturally hear. Just like how our sample rate and bit depth cover the resolution of human hearing. Of course, that bar will be raised once we invent cybernetic cochlear implants that not just restore hearing, but expand it. There's all sorts of fun directions that this discussion can go.
same goes with sight or smell. With cybernetic enhancements, more wavelengths (frequencies) would be visually detectable by humans, now this could either mean new colors, or something completly different that we can't even begin to imagine. And for smell, we could smell certain substances (like gasoline, which is, I'm pretty sure, odorless) instead of adding smelly chemicals to the substance.
I'd love to see a video showing the process of you creating that cello sound if that's at all possible? Love the channel dude!
I'd love to see that too
Explained so well! thanks for the video
Thank you for this video, you've inspired a project that I am pursuing for my Computer Science thesis project. I just got approval from the department today to design an fpga based additive synthesiser. I'll post all about it when I can, but it's about a year from full realization.
What happened? Did you do it? I've been waiting for 2 years! Haha
That's so cool, so how did it go
@@woulg Yeah lmao
This was unbelievable. Awesome explanation, and an amazing implementation.
I have no understanding of what I'm listening to but please keep making more.
This is the most impressive thing I've seen on the internet in the last week. Thank you.
As someone without a technical degree or a sound engineering background, just getting into music production, this video honestly shat on my ego and made me want to run in the other direction. Watching it a second time, I really appreciate the complexity, effort, and time that went into making this skillful demonstration and definitely want to start looking into additive synthesis and some of the terminology he covers on my own, so I don’t feel so retarded next time.
Would love to see a followup video demonstrating more about that cello resynthesis!
The Kawai K5000 was a hardware synth released in 1996. It could do a lot of similar stuff to what is demonstrated here. I have a Cello patch similar in terms of realism, will get around to sharing my patch soon to other K5000 owners.
that cello transforming between the realistic sound into the various synths and back again was sooo cool!
I hope you end up doing a part 2 to this, as it is a very fascinating subject. Covering how you did the cello and beyond would be fun to listen to (though I don't know if it'd be fun for your to make). Cheers.
Really dig that Cello randomisation, the way it alludes to the truth behind the sound's construction
I never realized that you liked analysis/resynthesis. Me too! But man you are so great at explaining and doing the math and all. 😍
This is the first video I've seen on your channel, it's incredibly well structured and researched, I subbed straight away
Cool to see Fourier analysis taught this way. As a scientist who deals with these algorithms, I mostly approve lol
This is incredible, basically with the correct maths you can make any sound that you can think of, and you can make incredibly detailed instruments without have massive sample banks, similar to your piano tec video, or is my understanding wrong
Even though synthesizing could require less space in disk since it only require parameters for the sound, the process of synthesizing is heavy for processors, that is why some digital/computer synthesizers usually create a temporary library with all the notes/pitches. Like a pre synthesis of the sound, so it reduce amount required process for rendering and live play.
Nice vídeo, thanks for sharing!
Wait, did you _create_ that cello in Starlight? Was that a synthesized cello that you created?? How on earth did you model the bow-on-string attack and stuff?? WHAT AM I LISTENING TO
I think he mentions swapping. So obviously there is the sample and then it crossfades between it and the modelled version, which sounds mangled. Not sure if that's correct, but sounds that way to me.
@@EvilDragon666 I got the impression, both from what he says and the video caption, that the cello is entirely additive. I think he might have used a sample and some sort of machine learning to get it spot on.
@@EvilDragon666 Nope, that's definitely not crossfading. He also mentions he starts randomising the partials that make up 'what you thought was a cello'. Seems pretty clear to me. TBH if the modelled sound was that far off why would he demo it as it'd disprove his entire argument.
likely just adjusting partials with envelopes. the Harmor plugin in FL Studio is an additive synth plugin that i believe lets you do things like this, but it sounds like he's pretty in tune with low-level synth programming stuff so it's hard telling what tools he used for it.
I actually tried resynthesizing a violin in Alchemy, by just importing a random violin piece that was in no way prepared for that, and Alchemy made an additive synth patch that sounded very close to the violin. I believe one could prepare a sample in a certain way for the additive engine to have an easier time analysing and recreating it, so these things are real. I easily believe this was an additive cello.
Where is your video covering the noise portion of the sound?
Its pretty much necessary if you want to make flute sounds
"Or is it" I fully expected the V Sauce theme to kick in
These videos are so great, man. I'm pretty stupid but still followed. Up until deep learning making sampling obsolete, but that's neither here nor there. I would love to have a video dedicated to that cello sound because my mind is seriously in pieces right now. Actually it'd be cool for you to do little breakdown videos of other things you've used in your music too. Just so we have real world applications in a creative setting- and then an explanation of how you got there.
Hi Ben. Great video. I fucking love your channel.
I disagree about "made sampling obsolete" however. I would argue that sampling isn't just about capturing the tone of an instrument. I think this makes instrument sample libraries a thing of the past, but actual sampling, as is the case in a lot of hip hop, for example, is about taking little chunks of a performance, usually recorded with gear you wouldn't have access to, or in a way you personally wouldn't. Sampling can be about creating a musical collage, rather than just digitising an instrument.
I dropped out less than a year before I could finish my math degree so I’m glad I have practical use for Fourier analysis now.
Wendy Carlos was so ahead of her time. She started using additive synths in the 80’s. I believe Tron was done with an additive synth.
wow...I love the sound you make...I LOVE that album..!
Oh wow, what a gem to hear the isolated track from Piety and the method to which you created it.
I work in Linux, and recently I've gotten obsessed with experimenting with the additive synth "Organic", which is an eight channel additive synth in LMMS. I feel like, like everything else, this is taking me down a road.
I still prefer my vintage analog additive synth from the 1960s. It has nine harmonic sine oscillators called ‘tonewheels’ and full polyphony.
Forgive my ignorance but I'm just now realizing I'm watching THEE flashbulb! The lawn wake series got me through a lot of hard times. Thanks not only for the music, but for the education too
Ummm I'm a cellist, and i was like oh great, here comes another weak cello imitati- 11:41
when the pitch bend of a hard cello bowing came through, I was beyond shook
I don't know whether to be offended or amazed,
I chose fearfully amazed🙏
As soon as you mentioned Machine Learning my mind went in to infinity 🧠
Dude, thank you for the time you put into this to explain it to us.
I get it much more now, the small amount of time I spent with additive synths made me think they were somewhat limited, I was so very wrong!
i love the 3d graphics! great production value
You're really a sound alchemist!! Such an amazing videos. Thanks!
dude. this is like. the best video i've ever seen.
Not a musician, just a DJ but absolutely in love with your channel. Thank you so much for all your educative work!! It really means a lot to get input from down to earth people who embrace science !!!
DJs mix stuff that they didn't even make at parties, events, they don't program synthesizers to play unique instruments and make songs. The flashbulb is definitely a musician.
@@JiMMy-xd8nu I was refering to myself (the dj) and just praising the musician (benn Jordan) for his amazing educative work. (as a DJ I highly value musicians dedication and passion for music creation).
@@goldfish3385 Oh I completely missunderstood your comment, that's my bad.
@@JiMMy-xd8nu no worries, my english isnt tip top, so can easely leed to confusion
@@goldfish3385 Hey, my english isn't the best either, and your comment wasn't that confusing, I just stopped functioning while reading it the first time.
Pleaaaaase do more videos about additive synthesis and modelling! The track with the chello is bonkers! I really want to know how to do things like this
Outstanding video, thank you !
I am studying csound in my free time in order to program cool synthesizers. I've got some pretty good results so far, but I hope there will be a video on csound on your channel one day.
You're running through this sooo fast! Yet, I've found it incredible useful 👍 really enjoying Piety of Ashes now 🤖
I'd like to thank all those mathematicians & engineers whose insight have vested on us a way to understand how systems work.
Moreover, all these transforms are all related... Even on quantum mechanics!
And Ben, you've done an outstanding job. Thank you.
this was fantastic, thank you im a uni student and have chosen additive synthesis as my main topic for an online presentation. this helped a lot. thanks dude!
There's a new VST synth that builds it's sounds using nothing but sine waves, and a lot of clever modulation. Sines, lovely!
Really an inspiring introduction to additive synthesis!
3 years later and I'm curious how far additive stuff has come.
not gonna lie. those last 50 seconds was simply mindblowing and inspiring!
Fascinating video, definitely subbed, and at this point in time, I'm seeking it video content like these to really become better at sound design, and understanding how to make anything from basically nothing 🤔👀 thanks for this video tho, learned a lot
Brilliant video Benn. Loved it.
I didn't know you were the flashbulb, I've never sat through the entirety of red extensions of me, but sensual data is one of my favorite songs ever.
I really wish you were correct in being confident about the future, 6 years onwards... I'm still looking for a harmor equivalent that does resynthesis that's not daw locked ;__;
When describing the basics of additive synthesis, it sounds just like using draw bars on a Hammond organ! And your first example of doing a few harmonics sounded just like a B3. But, because the Hammond used physical sine wave generators (tone wheels), it could only be built with just under 100 of them, and their frequencies are tempered, such that the harmonics are shared across every note, blending across all notes played, and especially blended when pulling all the bars out. Then it becomes pretty hard to achieve dissonance and instead there’s a smoothed out harmony of every note played to each other. Great for rock and roll players to just go wild, never a false note, but of course never a true one either (hence cheesiness). I don’t think samplers capture this effect, and I wonder if there are true circuit emulators that do (I don’t follow modern gear enough to know). But your mathematical breakdown of the unique harmonic series frequencies for every note one would need to reproduce things accurately is a nice clarification of how challenging this kind of tech would have been to consider in the time of early electronic and electro-mechanical organs!
Readditive synthesis is in sample based music is the future, i fucking love it.
Very nice introduction and example about Additive synthesis
So where's part 2? What synth were you using for the Cello? Great videos by the way.
Cats don't bark. Entire video invalidated.
Hi, I really like the way you are explaining things in your videos Benn! I am actually trying to create additive synthesis for my master project, so I am really looking forward to seeing the next level tutorial... We can find some almost similar explanations here and there, but one always lack of deeper understanding for going further, in order to reach more realistic results... I am not talking of perfect or high level recreation, but more how to enter in the real world of additive synthesis after the basics... Hope you'll find the time for us man, and long live your channel! Greg
OH MY GOODNESS!! the cello business had me floored. just amazing man
This is encredible. What the heckeroonies.
Sine waves adding up to look like different things reminds me of string theory
wow, amazing, I'm looking up your track now, it sounds great. thanks for the video :)
Woah, that was mind blowing. Awesome video, loved it!!
Benn, Just friggin Awesome! Looked you up on Spotify. Holy carp! Awesome!!!!Awesome!!!Awesome!!!
mate you're an incredible wealth of knowledge! thank you
Additive synthesis is also the past of Synthesis. When You made Your dull saw You made a mistake, it was instead a "sharp" saw because You neglected to reduce each successive higher harmonic amplitude by 1 over it's frequency.
Video quality is great!
The Hammond Organ is like that too
I'm not exaggerating, when you said:"combined with deep learning", I jumped around and cheered for 15 seconds. From now on, I'm gonna start recommending this channel to ppl.
Thank you for clearing that up. Useful info.
Utter dope ness merci boontjes
What if you had essentially 128 resonators for each key of the piano. Essentially each key is a very sharp EQ correlating to the note. You could also put a gate on each key and a trigger so the amplitude of the individual resonator notes would trigger MIDI on a synth. You could also do things like have each resonator's pitch stretched or arranged in a microtuning so when audio is fed through it, only harmonics relating to the current tuning of this resonator synth can pass through.
When you mention that additive synthesis can be used where disk or memory resources are limited (like games on microprocessors etc.), it sounds like you are reinventing lossy compression (like MPEG/Audio).
I didn't know the flashbulb was making videos, wow, just wow!
I can't understand what you saying but I truly appreciate this video and will be re-watching this to try and grasp each thing that you are saying, because I just started learning and designing synthesis and for some reason I would subtract everything until I got the frequency range that wanted, regardless of dB and just say to my self, I wish I could layer over or bulk up the dB of this range because I have the sounds sculpted to the way I wanted. Kind of a novice, dumb, and borderline unintelligible comment, but the entire video I was saying: I don't understand what you are saying, but to a degree, I get what you are doing, I'm just not formally educated on the matter. Thanks for this! Hopefully I can grasp some of these theories and notions sometime soon so I can explain what I already kind of knew to others! The sine is pretty much the water of sound haha
I kinda understood this, thanks for the video :D
I love jamiroquai! nice! my third album i ever bought was the album with virtual insanity on it.... i believe it is called Supersonic?
been a fan every since and love all their other stuff
Great intro to additive synthesis, very educational (even if i don't completely understand everything)! Loving your content so far! How would you program this in Csound?
Check out LayR synth. It is basically an almost unlimited number of subtractive synthesizers that can be added together making additive synths with more complex waveforms. For instance when you make a bell tone in LayR it is easy to create waves for each overtone. Very cool 👀
I'm nerdgasming over these videos.
As far as I understand is the basic principle of FM Synthesis also to stag layers of sine waves ( especially the feedback operator ) to get such results... Isn't it also a additive synthesis?
I've get myself my first FMsynth while waiting for my Tracker and - I admit quite confusing in the begining - but I guess there's a lot potential in it
FM achieves the same result but with different (and I would say more efficient) means. When you modulate one sine wave with a different sine wave as is done in FM synthesis, the resulting waveform is basically a sine wave made of another sine wave. Imagine the shape of a sine wave, then zoom out a little and imagine bending that wave into the shape of a bigger sine wave. The resulting sound is the same fundamental pitch but has more overtones (aka partials) giving it a more vibrant timbre. You could create the exact same timbre by adding the right sine waves together but it would require a lot more sine waves, depending on the ratio used in the FM synthesis. Basically both FM and additive synthesis could theoretically perfectly replicate any complex sound, they do it in different waves. I think of additive as adding sines together and FM as multiplying sines together
Amazing video - great content and well explained.
I'd love to see part 2 of this!
Wow! I had no idea of the power of additive synth until this video. I'm now on a mission to emulate a clavinet this way. You said your demonstratrion of capturing the melodica's timbre is "not the most efficient way of doing additive synthesis in 2017." What is a more efficient process?
My thoughts are - additive synthesis is definitely a useful tool, but it is never going to make samples obsolete. Just IMO! You need a ridiculous amount of processing power to handle enough partials (and allow good enough polyphony!) to completely fool us. To emulate a static sound is not that difficult and I mean you can do it in Alchemy (700 partials max). But the big kicker is emulating the actual physics and behavior of a particular instrument (which is easier done with physical modeling than additive synthesis), THAT would take an awful lot of time to get right for each instrument. It took Pianoteq 10 years to get to an acceptable level of quality (and that's not an additive synth - so for additive synth it would be even more difficult/time intensive!). And what about the recording chain? The mic, the room? Samples still portray this most faithfully, and you cannot reproduce this with 1000 partials. Maybe 10000 would suffice.
And ultimately: sampling is SIMPLE to use. Additive synthesis is anything but. Yeah you can get to places with machine learning, but that is still not really layman people territory (programming such algorithms), while anyone can grab a mic and a recorder and... do stuff with it.
I say all this as a supporter of physical modeling (I absolutely adore Pianoteq), but also with deep knowledge about sampling (since I've been developing Kontakt libraries for the past 13 years).
11:40 great track
You should make a video on how you made that cello!
super keen but cant find that follow up additive synthesis video on your channel....
Excellent video , really helped me to understand this synth business better :D ty
Wow. Not easy for me to follow, but the content is great and the video is very well done. Thanks. I'd be interested in physical modeling and eurorack patching :)
these motion graphics are next level
Excellent content, thanks so much.
Blew my freakin' mind.
Fourier series rule 🤓 awesome explanation and demonstration.
Not sure if you have used the Make Noise telharmonic, but it is a pretty cool additive synthesizer. It doesn’t have 500 sine waves and can’t do what you did there at the end of the track, but it can recreate lots of sounds well.
Mind Blending! Thank you a lot!
hopefully you'll see this, was that final sound achieved in either Alchemy, Csound, or Kyma? they are the only two systems I could understand such a sound being made in... harmor and morphine dont seem like they could do such a smooth "randomisation" of the partials.
This is brilliant.
Hartmann Neuron synth, we need to bring it to PC! If you haven't heard of it open an new tab and look it up. It's the saddest loss in the synth industry in my opinion. Work is being done to bring it to PC but I think it should be given new life.
This was such a sick explanation mate, so fuckin cool
Fuckinell Ben la, you're blowing my fuckin 'ed off here mate.