What Are Formants?

[dannydawiz]

Hey guys. I've been getting a bit more into dubstep sound design and this term keeps on popping up. Is it important to understand? What is it exactly? Also... I've been finding it really hard to design these sounds using just sylenth. Is massive better for programming dubstep sounds in general?

[bandcoach]

formants are probably better described as instrument/vocal tract resonances that do not change despite the pitch changing - all notes on an instrument have the same basic harmonic structure, but these are further enhanced by the natural resonances of the instrument body so that some of the harmonics within the harmonic structure of the note stand out more than others

think of it as tunable resonance filters applied to a range of frequencies within the audible range of human hearing

[dannydawiz]

formants are probably better described as instrument/vocal tract resonances that do not change despite the pitch changing - all notes on an instrument have the same basic harmonic structure, but these are further enhanced by the natural resonances of the instrument body so that some of the harmonics within the harmonic structure of the note stand out more than others

think of it as tunable resonance filters applied to a range of frequencies within the audible range of human hearing

I see... Does that mean that no matter what note you play, the formant of an instrument will always remain the same? Also... I usually see the term "formant" used when it comes to vowel sounds as well. Do you know how the two terms are related?

[bandcoach]

as stated before a formant is the natural resonance of the generating medium - it can be the instrument body or the vocal tract; which has several formants based on nasal, throat and mouth cavities

these cavities produce consistent resonant freqs when someone speaks and this is why it is easy to recognise different peoples voices and why we can also categorise different vocal types (amongst things that help with that identification)

as for formants in instruments the resonances remain the same regardless of what note is played - it is why some notes appear to be produced effortlessly (they sing out from the instrument) whilst others are difficult to produce (they seem thin and weak)

[scrapheaper]

Sylenth is a bit of beginners synth imo. It does cover the basics very well, but a lot of people find it limiting as it doesn't do FM, waveshaping/wavetable or sample based synthesis. Massive is one option for a general purpose synth with a few more bells and whistles, but don't think it's the only option- I personally think serum, synthmaster and z3ta 2+ are just as fully featured (if not more) and cost less money.

The classic (and probably overused) dubstep growls are usually made by FM synthesis. Massive can do this, but it's a bit limited and a specialist FM synth such as FM8, Toxic biohazard is much more useful- have a look at the seamlessr tutorials. There are some pretty powerful FM stock synths which are definitely capable of making formant-y growls- if you own FL studio or Ableton you could definitely use operator or sytrus to make such a sound.

In terms of how formants are related to vowel sounds, you are correct in saying that formants don't change with the pitch of the sound. Vowel sounds are created by specific sets of frequencies of formants- I believe the wikipedia page on formants can tell you which exact frequencies you need to make the an 'a' an 'ee' an 'oo' etc.

When designing sounds to sound vowelly, here are three methods I use. One is to just play around with synthesis until you happen to hit on something vowelly by chance- FM synthesis often does this, so does sample rate reduction and certain combinations or resonant filters.

A more deliberate approach would be to use vocoding or a formant filter (I like forma-8, especially as it's free), which deliberately introduce formants that are the same as the ones generated by a human voice.

[dannydawiz]

Sylenth is a bit of beginners synth imo. It does cover the basics very well, but a lot of people find it limiting as it doesn't do FM, waveshaping/wavetable or sample based synthesis. Massive is one option for a general purpose synth with a few more bells and whistles, but don't think it's the only option- I personally think serum, synthmaster and z3ta 2+ are just as fully featured (if not more) and cost less money.

The classic (and probably overused) dubstep growls are usually made by FM synthesis. Massive can do this, but it's a bit limited and a specialist FM synth such as FM8, Toxic biohazard is much more useful- have a look at the seamlessr tutorials. There are some pretty powerful FM stock synths which are definitely capable of making formant-y growls- if you own FL studio or Ableton you could definitely use operator or sytrus to make such a sound.

In terms of how formants are related to vowel sounds, you are correct in saying that formants don't change with the pitch of the sound. Vowel sounds are created by specific sets of frequencies of formants- I believe the wikipedia page on formants can tell you which exact frequencies you need to make the an 'a' an 'ee' an 'oo' etc.

When designing sounds to sound vowelly, here are three methods I use. One is to just play around with synthesis until you happen to hit on something vowelly by chance- FM synthesis often does this, so does sample rate reduction and certain combinations or resonant filters.

A more deliberate approach would be to use vocoding or a formant filter (I like forma-8, especially as it's free), which deliberately introduce formants that are the same as the ones generated by a human voice.

This post was very informative. I actually produce on Logic Pro and not FL Studio. Sylenth1 does have a few lfos that you can use for FM synthesis. The main thing that I like about it is that it's really simple in the sense that most sounds are either made using a saw, square, or sine wave.

The thing that I find tough about dubstep sound design is the fact that it isn't as simple most other genres of EDM. All of a sudden you have these sounds that are distorted to hell and are being modulated in 2-3 different places.


I've been wanting to switch to massive recently but the only thing I don't like about it is that it has a TON of different wavetables. I feel like the simplicity becomes compromised but then again I do feel like Massive is easier to work with at least in regards to FM synthesis. If I make a sound I want to understand how make it conceptually straight from a saw wave or square wave. I don't like how massive has all of these "modern talking" etc... wavetables that I don't understand how to make starting from just a simple saw wave.

Then again I may check out the FM8 since It does seem like something a bit more new and exciting.

[steffeeh]

If you want to make a formant filtered sound, you need a synth where you can setup 2 parallell filters - one with a bandpass, and one with a bandcut (this one is most important). Then you control the bandcut width and the bandpass cutoff with whatever you like, and envelope, LFO, automation etc. Experiment with the knob settings on these parameters as well as the resonance on both of them to achieve the right movement.
The main formant sounds comes from 2 resonances in the frequency spectrum moving toward eachother, at least with the typical YEAY-sound. Hence why the bandcut filter is most important as you'll have 2 resonances moving as you control the bandcut width. The bandpass is more about balancing the sound.
Massive can do this, and Sylenth should be able to do this as well, however I don't use Sylenth, but I believe it has parallell filters.
Otherwise you could just do this OTB, using 2 parallell mixer channels with different EQ settings, or if you have a plugin where you can do parallell filters.
Anyway, I remember I've once stumbled upon a cheat sheet with what frequency parts that needs to be attenuated for what formant sound, so google it and see if you can find it (if you don't, let me know and I'll send you it, as I have it written down inside my DAW).

However I wouldn't recommend using formant filtering to create growl sounds... because they mostly sound really, really horrible and amatuerish, at least if you're going for a huge bass sound. But for funky pads and leads and such, or noisy effects with a more spicy shape on it, this technique can be perfect.
As for FM synthesis, it's a very recommended and a very reliable synthesis technique for this, but it's not the technique that produces the cleanest and smoothest growl sounds (like the ones Skrillex has), but more rough and naughty ones. Depends on what you want.

Check out this tutorial on a very advanced way to create very smooth sounding formant sounds:
How to actually make your own Skrillex Growls -


PS. "being modulated in 2-3 places"... that's cute

[bandcoach]

I usually see the term "formant" used when it comes to vowel sounds as well. Do you know how the two terms are related?

sorry totally missed this part of your question before

when we say words we change the resonant frequency of the mouth by changing the overall shape: different vowels have different shapes throughout the vocal tract and therefore different resonances - we use formant to describe the way in which the vowel is formed by the change in vocal tract shape

[scrapheaper]

The main thing that I like about it is that it's really simple in the sense that most sounds are either made using a saw, square, or sine wave.

FM synthesis uses combinations of sines, squares, saws and triangle waves to make more complex waves.

The thing that I find tough about dubstep sound design is the fact that it isn't as simple most other genres of EDM. All of a sudden you have these sounds that are distorted to hell and are being modulated in 2-3 different places.

As you learn more sound design you might find that this isn't uncommon. 2-3 modulations is still fairly basic compared to some sounds which have dozens.

I've been wanting to switch to massive recently but the only thing I don't like about it is that it has a TON of different wavetables. I feel like the simplicity becomes compromised but then again I do feel like Massive is easier to work with at least in regards to FM synthesis. If I make a sound I want to understand how make it conceptually straight from a saw wave or square wave. I don't like how massive has all of these "modern talking" etc... wavetables that I don't understand how to make starting from just a simple saw wave.

A lot of the more advanced sound design techniques often involve using samples as a starting point- any noise can be integrated into bass for a bit of organic-sounding filthiness. Have a look sample based synths such as alchemy, omnisphere, harmor, iris etc.

Also, a lot of the massive wavetables probably could be made by FM synthesis if you knew how.

If you do want to create your own wavetables from scratch rather than using massive's pre-made ones, have a look at serum- it's one of the most powerful and fully featured VSTs available today.

[dannydawiz]

Thank you so much for all the replies everyone. I'm sorry for being a bit late. School, homework, and work are starting to make free time a lot smaller recently.

I'm sure I can find that formant cheatsheet somewhere online. The funny thing is I just saw a tutorial featuring a couple of the monstercat producers (virtual riot, au5) talk about how they make those vowelly sounds. It looked VERY similar to what you're describing. Except instead of using a band reject they used two bandpass filters and automated the cutoff of one of them to "crossover" into the other.

I'm really excited to experiment with the tips you just gave me. Same goes to you band coach. I still have a lot of stuff that i need to go over from your other posts as well. What you said about the vocal Tract changing shape was quite informative.

When you refer to "parallel" eqs... What exactly does that mean? I really like the idea of being able to create this formant sounds just by using two EQs and automation. When it comes to sound design I feel like understanding the theory behind everything is far more important because once you learn it you can recreate it in essentially ANY synthesizers.

What is faster though do you think when it comes to FM synthesis? Working with the LFOs in the synthesizer or controlling things based on automation?


I'm trying to wrap my head around the theory behind FM synthesis. Is it really true that you can create any of the tables in massive using normal FM synthesis? If so... this makes me happy to hear. I'll take a look at serum once I get the chance. I really do feel like sylenth has finally reached its limits. At least until this stuff starts to become second nature.

How is it that a sound could have 12 modulation a going on at once and still retain its sound? I would assume that it would just turn into a big pile of mush. Then again... I'm not very familiar with FM synthesis at all aside from the basic "wobble" bass and the good old "YAI". I don't own a sample synthesizer either so I have absolutely 0 experience when it comes to sample based design.

I'll check out that video as soon on growls as soon as I get the time. Once again, thanks everyone for being so helpful.