How to mimic acoustic instruments with synthesizers - a cry for help!

[pekadan]

Nobody mentioned physical modelling yet? I'm sure if you followed up on @fmr's suggestion you'd come across PM material pretty quickly as Risset used and wrote about it pretty extensively I believe

I didn't look much into PM yet to be honest, but I probably should. I've played around a little with some of the NUSofting and AAS PM synths, so it's not completely unfamiliar to me, but I've got loads to learn for sure.

Can't seem to find Risset's book anywhere unfortunately. Neither on PDF or paper form. Same for Massey's book. If anyone has tips on where to get them I'd be grateful.

[himalaya]

Nobody mentioned physical modelling yet? I'm sure if you followed up on @fmr's suggestion you'd come across PM material pretty quickly as Risset used and wrote about it pretty extensively I believe

Myself, I didn't mention it, since it's too obvious, too easy, if you know what I mean. It's easy to recommend something like AAS String Studio for all manner of realistic sounds, but there is no fun in that. Besides, and I think this is more important, using such pre-fab modules like in the String Studio and other similar synths does not teach the fundamentals of synthesis, how to use basic tools like standard oscillators, filters, distortion, etc. in order to obtain such imitative sounds. And I'm saying this as a total fanatic of Physical Modelling, it's just, I don't think it's the answer in context of somebody learning and wanting to come up with 'acoustic' sounds synthesised from ground up, if that was the aim.

Mind you, some of the tools I mentioned, like comb filters, touch upon Physical Modelling and are used in order to imitate Physical Modelling, but here again, using one or several comb filters requires good listening skills in order to tune them properly. So this is a much better learning opportunity, than when switching through pre-fab body sizes of the guitar model in String Studio, as an example.

[vurt]

Now I'm off to start browsing the Sound on Sound articles about synthesis. Sounds like a goldmine!

theyre an amazing resource

[himalaya]

Wow, that's very interesting and useful info! It takes time and effort to get good at this, that's for sure, but I find your suggestions both inspiring and motivating. Thanks a million!

It's my pleasure I'm happy to talk about this. It's a great subject, very close to a lot of recent projects I was involved in.

The sound you demoed previously is amazing. Not only does it sound like a real acoustic instrument as far as timbre goes, but the way you're able to play it too, with articulation, sounds very "real". Incredible. My first thought was to ask you what synth you've used to get a sound like this, but reading your last post I realize that's not a very relevant question. (but when you release an expansion pack of patches for that synth that includes sounds like this gorgeous "flutuk", that's when I have to know, ok!? ).

You mention another very important aspect of imitative synthesis: realtime articulation. This is also crucial to get right. This is all about realtime control, so it refers to modulation, but it also may require looking at suitable midi controllers so that we can mimic certain gestures accurately.
The above example uses note legato pitch bends, transitions from note to note by gliding into each note. It also uses continuous pressure in order to get the 'tremolo' effect (traditional aftertouch can not accomplish this). This is all done in realtime, there are no LFOs doing any cyclic-modulation, and no pitch wheels... Explore instruments like the Haken Continuum, Roli Seaboard Rise, Roger Linn Linnstrument, or Madrona Labs Soundplane. These will allow you to get realtime gestures which are simply impossible to get on a traditional keyboard.

I'm curious to know what your approach to the attack of your woodwind instrument was. As previously said, it's often difficult to mimic the attack of the acoustic instrument with its complex inharmonic transients. Did you put special effort into getting the attack part to sound realistic? If so, will you share some thoughts?

Good question! In the above example, I started with a slow attack on the main amp envelope and slow attack on one of the mod envelopes. Then the attack stage of the mod envelope is controlled via velocity, so low velocity = slow attack, high velocity = fast attack. This is a great technique and I use it even when making analog-style sounds, which traditionally, are not associated with such modulation. However, I just love subtle attack modulation on many sounds that on the surface may not require it.

Velocity also controls a few other parameters in this way (LPF cutoff, and noise level) but the main attack actually comes from all the parameters assigned to ..... pressure, and coupled with the continuous pressure response on my midi controller, I can control the attack in a slightly different way to velocity. In fact, here, this continuous pressure response behaves a little bit like velocity, in that it can be used to fool the ear that I'm playing with velocity, but in fact, something else is happening. To illustrate it more, I could delete all velocity modulation, and I could still get what resembles velocity when I play softly and then hard, with force. The continuous pressure response will create an illusion of velocity modulation. It's a weird thing to explain, but it works! and can be used creatively.

[himalaya]

(but when you release an expansion pack of patches for that synth that includes sounds like this gorgeous "flutuk", that's when I have to know, ok!? ).

Hehe this particular sound is for something very special. It won't be released by me but another company.

However, I am planning to make and release similar sounds for a very popular synth, one which I feel has a great collection of tools, which should help to make a nice set of hybrid-acoustic sounds.

BTW, 'Flutuk' great name for a hybrid preset!

[Scrubbing Monkeys]

This is a deep and complex subject, and since you are late with your report, I fear not much can be done to help.

Subtractive synthesis can be a very good source of acoustic sounds if coupled with a flexible, modular signal path that offers wave-shaping, comb + formant filters. For example, you can create a very nice 'nylon' quitar pluck with a pulse waveform, comb filter and a LPF with a suitable filter envelope shape (to create a 'pluck' response).

Then, using a triangle wave through a waveshaper (with intensity assigned to a controller) into a comb and a LPF can give all manner of flute sounds. However, for this to be believable, you would need to have a separate source of noise going into one, or better, several comb filters (plus LPF or BPF) in order to simulate the breath noise. White/pink noise through a LPF or BPF alone is not enough.

Moving on from a modular subtractive engine, I'd recommend FM, above additive. I found FM to be much quicker to use and quicker to obtain great harmonic control. In fact, I have been making some acoustic sounds in a synth just in the past few days, an engine which can be described as 'analog subtractive' but one which also has very flexible FM capabilities. Here, I was able to get the basic tone just with the subtractive part of the synth, only adding overtones and more 'body' via FM. The FM is really only between 2 oscillators. The result is what sounds like a mix between a flute and a duduk or some other woodwind instrument. An imaginary acoustic sound, but to me it definitely sounds 'real'. I've uploaded a quick preview for you to have a listen:

https://soundcloud.com/electric-himalay ... ind-thingy

Very nice sound. It almost has a violin-esk attack at the stronger velocities. You input to this thread is quite inspiring. Thank you.

[pekadan]

You mention another very important aspect of imitative synthesis: realtime articulation. This is also crucial to get right. This is all about realtime control, so it refers to modulation, but it also may require looking at suitable midi controllers so that we can mimic certain gestures accurately.
The above example uses note legato pitch bends, transitions from note to note by gliding into each note. It also uses continuous pressure in order to get the 'tremolo' effect (traditional aftertouch can not accomplish this). This is all done in realtime, there are no LFOs doing any cyclic-modulation, and no pitch wheels... Explore instruments like the Haken Continuum, Roli Seaboard Rise, Roger Linn Linnstrument, or Madrona Labs Soundplane. These will allow you to get realtime gestures which are simply impossible to get on a traditional keyboard.

Cool! I think I will dedicate a part of my report to realtime articulation. I realise (especially when I listen to your demo) how crucial this is to imitating acoustic instruments, and how important it is for creating patches that are musically useful.

I'm curious to know what your approach to the attack of your woodwind instrument was. As previously said, it's often difficult to mimic the attack of the acoustic instrument with its complex inharmonic transients. Did you put special effort into getting the attack part to sound realistic? If so, will you share some thoughts?

Good question! In the above example, I started with a slow attack on the main amp envelope and slow attack on one of the mod envelopes. Then the attack stage of the mod envelope is controlled via velocity, so low velocity = slow attack, high velocity = fast attack. This is a great technique and I use it even when making analog-style sounds, which traditionally, are not associated with such modulation. However, I just love subtle attack modulation on many sounds that on the surface may not require it.

Velocity also controls a few other parameters in this way (LPF cutoff, and noise level) but the main attack actually comes from all the parameters assigned to ..... pressure, and coupled with the continuous pressure response on my midi controller, I can control the attack in a slightly different way to velocity. In fact, here, this continuous pressure response behaves a little bit like velocity, in that it can be used to fool the ear that I'm playing with velocity, but in fact, something else is happening. To illustrate it more, I could delete all velocity modulation, and I could still get what resembles velocity when I play softly and then hard, with force. The continuous pressure response will create an illusion of velocity modulation. It's a weird thing to explain, but it works! and can be used creatively.

Thanks a lot for the explanation! Very very interesting. I’m taking notes!

May I bother you with another, and what’s possibly a stupid, follow-up question? How does continuous pressure actually compare to aftertouch? I’ve been trying to understand this by reading various descriptions on the web about MIDI pressure, but I don’t really get it. It’s not velocity data at note on and it’s not pressure data sent after note on - but then what it is?

[Kinh]

I find it odd after reading these that no one bothered to ask him what acoustic instruments he's trying to model.

[himalaya]

May I bother you with another, and what’s possibly a stupid, follow-up question? How does continuous pressure actually compare to aftertouch? I’ve been trying to understand this by reading various descriptions on the web about MIDI pressure, but I don’t really get it. It’s not velocity data at note on and it’s not pressure data sent after note on - but then what it is?

Continous pressure is midi pressure data sent at note-on.

So in practive, as soon as you touch the surface, pressure level is scanned, and the deeper you press, the greater the effect.
This is obviously much different to how 'aftertouch' works. However, it is still possible to emulate 'aftertouch' if whatever software that you are using allows some kind of mod maps, or envelope shapers to be inserted between the mod source (pressure) and the mod destination (for example filter cutoff). Another name for these are 'expression curves'. And so by creating a deep exponential shape on such a 'curve' we get a response which acts much like 'aftertouch'. I'm highlighting this difference, since sometimes we want continuous-pressure (which works in tandem with velocity) and sometimes we want traditional aftertouch (which is separate to velocity, that is, it never gets in the way of velocity, like continuous-pressure can).

Mind you, the continuous pressure response can be obtained from other controllers. A mod wheel is a continous-type controller, or a pedal, or a macro fader, and these can give you a similar gestural response on monophonic lines, but not when it comes to polyphonic lines, since all those controllers are applied to all notes at the same time, whereas midi controllers sporting continuous-pressure normally do it polyphonically. Also, it has to be said that whatever that is available directly right under your finger tips is much more intuitive to use, so there are clear benefits of designs which use continous-pressure.

[pekadan]

I find it odd after reading these that no one bothered to ask him what acoustic instruments he's trying to model.

Well I’m not being purposely vague about that for some reason. It’s just that I’m interested in what himalaya describes as imitative synthesis in general and haven’t even decided yet what instuments I specifically want to try to imitate for my presentation. One of the failed experiments that prompted me to start this thread was trying to mimic an acoustic guitar and another one was the violin.

[pekadan]

May I bother you with another, and what’s possibly a stupid, follow-up question? How does continuous pressure actually compare to aftertouch? I’ve been trying to understand this by reading various descriptions on the web about MIDI pressure, but I don’t really get it. It’s not velocity data at note on and it’s not pressure data sent after note on - but then what it is?

Continous pressure is midi pressure data sent at note-on.

So in practive, as soon as you touch the surface, pressure level is scanned, and the deeper you press, the greater the effect.
This is obviously much different to how 'aftertouch' works. However, it is still possible to emulate 'aftertouch' if whatever software that you are using allows some kind of mod maps, or envelope shapers to be inserted between the mod source (pressure) and the mod destination (for example filter cutoff). Another name for these are 'expression curves'. And so by creating a deep exponential shape on such a 'curve' we get a response which acts much like 'aftertouch'. I'm highlighting this difference, since sometimes we want continuous-pressure (which works in tandem with velocity) and sometimes we want traditional aftertouch (which is separate to velocity, that is, it never gets in the way of velocity, like continuous-pressure can).

Mind you, the continuous pressure response can be obtained from other controllers. A mod wheel is a continous-type controller, or a pedal, or a macro fader, and these can give you a similar gestural response on monophonic lines, but not when it comes to polyphonic lines, since all those controllers are applied to all notes at the same time, whereas midi controllers sporting continuous-pressure normally do it polyphonically. Also, it has to be said that whatever that is available directly right under your finger tips is much more intuitive to use, so there are clear benefits of designs which use continous-pressure.

That’s an excellent explanation! (Even) I get it now!

I remember that earlier you stressed how important it is to “learn your synth”. What would you say are the most important features of a synth to be used as a workhorse for imitative synthesis? From what I’ve learned so far, a subtractive synth with FM capabilities is useful, an extensive effect section (for effects like combfilter) and the more MIDI and modulation options (e.g. expression curves), the better. Other things??

[himalaya]

The simple answer is, anything and everything that can add 'colour', anything that can alter the timbre. So, in addition to the above, wave-shapers are also very important. Different flavours of distortion, basically. These really come into their own in a modular signal path, since applying subtle distortion at various points adds to the overall timbre. Then, it's about controlling these in realtime via pressure or modwheel. This is for example essential for woodwinds, reeds...

Formant filters/Band-pass filters - these are very useful, especially in combination with comb filters. And by 'formant filters' I do not mean one filter that emulates a single human voice vowel. I'm referring to something more complex, a series of band-pass filters which can create more complex shapes than a dual-band pass, usually responsible for the standard vowels in many simple formant filters.
Having access to frequency, bandwidth and amount, is important. Of course, in a modular design, we could create a custom formant filter with several band-pass filters in series, but it's great to have one module with several bands and all controls on one 'page' since it's just nicer to edit then.

Wavetables - this is another very useful feature. Even in its basic 'static' form ie: static playback of a single-cycle wave. Having the option to choose from nice selection of single-cycle waves can yield great results. This is especially useful if you have waves derived from acoustic instruments. Taking this a step further, if the wavetable has a series of waves all derived from one acoustic instrument, with waves set out in a harmonic relationship, then this can be used dynamically, to scan through the waves and maybe add more interest to the final sound. This could be very useful when doing the wave scanning on pitch-bend gestures so that the harmonic movement created by the wavetable scanning mimics the harmonic change that normally occurs in real life (during pitch bend). This technique is complex to execute well, but it can work. It's also esoteric and I doubt that most people would even consider doing this.

A modular signal path is always very useful, or failing that a very flexible modulation system. However, having the ability to insert various modules at specific points in the signal chain is very helpful. Also, it's very important to be able to treat various oscillators and noise by separate filters, so if using a non-modular synth, having some routing options is must. Some traditional subtractive synths can do it, but modular designs are just simpler to use.

Different types of noise waveforms are useful to have, although, noise can be shaped with filters, so white noise could be all you need. However, in my experience, to get from white noise to, for example, breath-type noise, takes some effort and good comb filters(s) with LPFs or BPFs. Also, the noise element of a bowed string is very specific, and nobody seems to model these and offer them in synths. I think this is one area that has plenty of scope for development (if any devs are reading this).

Equalisation - very important. Often, when using comb filters, or FM, or ring-mod, EQ is needed to cut problematic frequency bands. At other times EQ can be used to create more formant peaks. I tend to use EQ in one specific synth to help highlight various frequency bands in realistic brass sounds, most notably trumpets. EQ is very useful here to create various 'jazz' style muted trumpets. Maybe, if I get the time I will dig out a few of those brass presets I made some time ago, and post a few examples...however these are in a different studio, so wait and see...

I purposefully do not mention resynthesis since like physical modelling, this is a short cut, and does not really teach about the fundamental aspects needed in imitative synthesis.

Then, there are very complex features like the Grain Silo oscillator in the Eagan Matrix synth engine (for the Haken Contiuum). This uses a form of granular synthesis. Very complex but ultimately, in the context of the whole synth engine capable of amazing results.

Having said all that...a list of features is not that helpful without the skill needed to use them. It's a cliché perhaps, but so relevant here. I think the previously mentioned Sound On Sound articles can shed some light on how and when to use some of those features.

[pekadan]

The simple answer is, anything and everything that can add 'colour', anything that can alter the timbre. So, in addition to the above, wave-shapers are also very important. Different flavours of distortion, basically. These really come into their own in a modular signal path, since applying subtle distortion at various points adds to the overall timbre. Then, it's about controlling these in realtime via pressure or modwheel. This is for example essential for woodwinds, reeds...

...

Wow, this is fantastic information! I am so grateful for your input himalaya!

I need to start putting the pieces of this puzzle together. Writing my report in parallel with putting my new knowledge to test by trying to come up with some useful patches is next. And I'm really excited about it!

Having said all that...a list of features is not that helpful without the skill needed to use them. It's a cliché perhaps, but so relevant here. I think the previously mentioned Sound On Sound articles can shed some light on how and when to use some of those features.

That makes perfect sense. I must admit I have been playing around with some of these features you've listed above (when checking out some new synths for example), but always in quite random ways without really understanding what they're good for, or how to use them properly. Now that I really have a purpose learning about them, the experimentation is going to be so much more interesting and rewarding!

[Straight2Vinyl]

Nobody mentioned physical modelling yet? I'm sure if you followed up on @fmr's suggestion you'd come across PM material pretty quickly as Risset used and wrote about it pretty extensively I believe

I didn't look much into PM yet to be honest, but I probably should. I've played around a little with some of the NUSofting and AAS PM synths, so it's not completely unfamiliar to me, but I've got loads to learn for sure.

Can't seem to find Risset's book anywhere unfortunately. Neither on PDF or paper form. Same for Massey's book. If anyone has tips on where to get them I'd be grateful.

Physical modeling is the way to go for the most realism, but I suspect it's more complex than most synth designers could manage. Personally I can't imagine ever being able to create something like the SWAM instruments I own no matter how long I worked at it and studied.