Ah... vocal formants!

"Now, back to formants!" - Explorer

The easiest way to explore formants these days is probably by using DSP on your PC.. If you have an audio package capable of taking VST (or any of the other) APPS, you can just download one of the many free apps and try them out on theremin sounds.

This is one of my favorites - it models almost exactly the sort of interface I was after

http://www.a-quest.com/products/vocalizer_en.html

The Csound manual is a great source of useful data - Here is a table of formant frequencies .. From this table you can see why having simple X/Y control over the formants, and 'mapping' these is areal challenge - particularly in analogue. Vowels etc are recognisable with just F1 and F2, but really you need at least F3 if you want more than just a 'vocal influence' and actually want something that sounds like a human voice.

I have given up trying to make an analogue electronic clone of a singer.. 'just a touch' of vocal effect is enough I think... One not only needs to alter the formant frequencies, but also their independen levels, dynamically, if one really wants correct synthesis - IMO though, manual control over 2 formants, with another "formant" which tracks the sum of these two (easy to implement if the filters are voltage controlled) can give a nice vocal effect.

For those looking at the Chatterbox circuit - modern opamps are fine (TLO72 etc) and if you replace the potentiometers with 10k resistors, and put the FET side of a H11F1 opto isolator across these resistors, you can control both resistors in each filter by the current you pass through the H11F1 LED's.. This was my first mod of the circuit (in fact, I dont think I ever made one using ganged pots).

Once you have found the raw sound you want, the biggest challenge in trying to emulate any instrument (or the human voice) is finding the right trigger device. For me, the problem with the Vocalizer VST in the demo above is that it is being triggered by a keyboard. Any single note, played alone, might sound convincing but a chain of notes in a melody (such as SOLVEIG'S SONG in the demo) sounds disjointed, like Robbie The Robot, because the keyboard is not connecting the notes the way a real singer would. 

Fred, am I correct in assuming that the player in the demo is controlling the formant with the toggle we see him manipulating in the lower left corner of the screen?

"Fred, am I correct in assuming that the player in the demo is controlling the formant with the toggle we see him manipulating in the lower left corner of the screen?" - Coalport

I was under the impression that this software requires an X-Y pad input - and presumed the keyboard had such a pad at the lowe left corner.. or perhaps a joystick or some pad equivalent.. I ran this app under Cakewalk a while ago and remember having some hassle getting it to recognise my pad- I managed to get it to run on a laptop with its pad and used it briefly just to explore vocal effects.

By "trigger device" I presume you mean something that dynamically alters the formants to produce vocal effects like vowel sounds etc.. I believe that most of these processors (TM and the like) use something from the input signal (amplitude change, pitch change, harmonic change or whatever combination) to automate or "trigger" formant movement... the player has no dynamic control over this "movement" and can only select some preset.

My experiments have all been focused on non-automated control using joystick or equivalent - with this, the audio is continuously fed to the filters which are manually controlled - there is no "trigger" mechanism.

I think perhaps that the above demo does have some "triggering" which gives stronger vocal charactaristics, but the player also has some real-time manual control of the formants.

If one is manually controlling the formants then they can follow their own pattern without being affected by the input signal - the player is in control of pitch, dynamics, and formants - exactly as one is if one is singing (particularly if the audio source is a theremin or other instrument without fixed intervals).. As to whether one can come close to realism, well, I doubt it - Using apendages to control all these simultaniously is unlikely to provide anything like the control we have in our evolved vocal system.

I do, however, strongly suspect that someone like you, who has the musical aptitude and understanding of how vocals "should" sound, would manage a far better rendition using some formant  filter app, than most people could manage - simply because you can hear what isnt "right".. I never noticed the "disjointed" vocals in the demo until you brought them to my notice - oh, I could hear they were imperfect - but they never bothered me... They (to me) are a lot "better" than anything I have heard from the TM.

By the way, thank you to all concerned with this forum and this thread. Implementing formant filters in Audulus is on my to-do list as part of a larger project - so it might be a while before I get to it, and being able to refer easily to this thread will make it a lot easier. :-)

"Once you have found the raw sound you want" - Coalport

This in itself is no simple matter - and is a grey area IMO where "Vocal qualities" versus "really sounding like a vocalist" comes into play..

The levels of harmonics in a vocal sound are critical to realism IMO - the biological "formant filters" recieve a waveform from the larynx which actually changes quite little, and the major components which make us sound human are how the filters are adjusted dynamically.. We can hold our formants as we change pitch (as in, the formant frequencies dont change, and the waveform from the larynx doesnt change much - we have much less control over the larynx waveform than we have over the other components).. As the frequency changes, the energy present at the formant frequencies changes.. and the spectrum emerging carries the emphasis on the formants.

If one does not have a 'larynx like' waveform going into your formant filters, then the incoming spectrum will be different, the amplitudes of harmonics 'boosted' by the formant filters will be different, and as pitch changes what comes out will still have emphasis at the vocal formants but the fundamental pitch versus the level of output harmonics will not be perfectly human.

The larynx produces a waveform with high harmonic content, both odd and even harmonics are present, and with some 'rougher' voices there are enharmonic and noise components - these excite the formant filters and add emphasis to the formant components ... Formants are quite low in frequency, and adding some LF noise when passing higher frequencies through them can enhance the vocal effect.. In fact, I found that producing a harmonically rich signal an octave below that being played, and adding a small quantity of this directly to the input of the filters (or modulating the input to the filters) gave stronger vocal effects..

But they are effects - they would never pass off as someone singing! ... Yeah - I had this dream of being able to create a theremin-based "clone" of a human singer.. It was an extremely silly idea!  ;-)

"being able to refer easily to this thread will make it a lot easier." - Gordon

LOL ;-) You hope.... Second to theremins, vocal synthesis is probably the next most quirky electromusic undertaking one can engage in, IMHO..

One tip - the easiest satisfactory exitation waveforms I have found are 1/2 wave rectified triangle or ramps.. Or at least these are the simplest to implement in analogue circuits.

I hate fighting uphill battles.

Ah, well... I thought I was signing up on the theremin discussion board, and was excited to join others who were pursuing the same, and maybe create that same excitement for those who would come later, looking for the same board I believed I had found.

I have a host of things I've been working with, specifically with the theremin, and was hoping to talk about the theremin being used with and through them. I don't have the energy to keep things on-topic when so many don't want the same though, and so I'll not contradict the established members.

It's a shame, though... I would have thought that something called Theremin World would have members who would want to nurture that kind of interest in the actual theremin, and thereby help the instrument grow as something other than a novelty.

I'll look elsewhere. Cheers!

Jason started the EFFECTS forum at explorer's request because the gentleman presumably wanted to discuss this "host of things" he says he has been applying to the theremin. I guess now we will never know what they are.....why do I get the feeling this is no great loss?

I find it strange that someone who plays a Gakken Premium toy theremin, and who seems primarily interested in the application of FX, is so concerned about the theremin growing into "something other than a novelty". 

It would seem to me that the kind of things that interest explorer are the very things that will ensure that the theremin remains a novelty forever. 

Wow. He enters a community as a new user. He suggests that a forum should be created. The forum is immediately created. Then he creates a post. He gets replies, including from some of the most prominent thereminists in the world. And he leaves, vocally complained that the thread went off-topic (when it actually didn't?)

This must be one of the most riduculously rude attitudes I've ever seen!

Wow. He enters a community as a new user. He suggests that a forum should be created. The forum is immediately created. Then he creates a post. He gets replies, including from some of the most prominent thereminists in the world. And he leaves, vocally complained that the thread went off-topic (when it actually didn't?)

This must be one of the most riduculously rude attitudes I've ever seen!