Manny’s FM-Xpert #1: “Acoustic Eccentricities and Stuff”
Manny’s FM-Xpert is a deep dive into the the FM-X engines of the MONTAGE and MODX Music Synthesizers. This five article series is delivered in an accessible and entertaining way and includes MONTAGE content programmed by Manny himself. The other articles in the series are accessible via the links below: Manny FM-Xpert 2: “I’m Fixing […]
Manny’s FM-Xpert is a deep dive into the the FM-X engines of the MONTAGE and MODX Music Synthesizers. This five article series is delivered in an accessible and entertaining way and includes MONTAGE content programmed by Manny himself. The other articles in the series are accessible via the links below:
Manny Fernandez has been involved with sound programming and synthesizer development for over 30 years. Initially self taught on an ARP Odyssey and Sequential Pro-One, he also studied academically on Buchla modular systems in the early 80’s. With a solid background in analog synthesis, he then dove into digital systems with release of the original DX7. Along with his aftermarket programming for Sound Source Unlimited, Manny is well known for his factory FM programming work on Yamaha’s DX7II, SY77, SY99, FS1R and DX200 as well as the VL1 and VL70 physical modeling synthesizers.
Note: Remember to watch the video at the end of this article!
In the pursuit of emulative synthesis – recreating traditional acoustic instrument sounds – often where pure synthesis has often come up short compared to sampling is in recreating what I like to refer to as the ‘acoustic eccentricities’ of a real instrument sound resulting from its physical properties, construction and imperfections. These sonic components have been traditionally referred to as “stuff” — the aperiodic noise components, inharmonic overtones and transients along with mechanical noises in acoustic sounds that are key to their character – think of the breath in the attack of a flute sound or the scrape of the bow on a cello string. Sampling is a very effective ‘brute force’ solution where we capture the stuff exactly. But other than filtering there’s little to no dynamic or realtime control of the timbre, having to rely on velocity switched multisampling essentially ‘fake it.’ Plus, stuff can vary widely in timbre and characteristics across the note range of an instrument so there’s all the note multisampling as well. At the end of the day, this is a somewhat crude way to get around the fact you’re playing a sound who’s harmonic behavior is ‘frozen’ as captured in the original recording of the sample. It’s not really ‘synthesis’, which for me means creating and manipulating harmonic structures dynamically and over time in the manner I choose. Physical modeling synthesis advanced the ability to build dynamically and realtime controllable “stuff” but is unfortunately very complex and time consuming to program to achieve that controllability in effective emulative synthesis.
In contrast, one of the great capabilities of FM synthesis is the ease and efficiency in creating “stuff.” The basics for this consists of creating waveforms with non-integer harmonic series by setting your Operator Frequency Ratios to non-whole number values and/or using Fixed Frequency Mode Operators. The most well known examples of stuff in FM synthesis is in many electric piano sounds– the clinks, clanks and metallic attack transients of the ‘tine’ portion of the sound. So by using a few Operators to make this stuff you can improve the character and detail in your synthesis. Thinking big, consider what can be possible in FM synthesis if you have a lot of Operators available. When Yamaha released the TX816 module way back in 1984, it was typically used as an 8 Part multi-timbral DX7. But in another context, is was a bunch of Operators in a box that could be used instead to create one single sound. Those who programmed the TX816 in that context have experienced what can be accomplished in the ‘old school’ DX7 FM engine when having 48 simple sine wave Operators available to build a single sound.
With Montage, we now have the FM-X synthesis engine that expands the Operator capabilities with non-sine waveforms and an implementation of Operator Fixed Frequency Mode that allows for variable key frequency scaling and velocity modulation of frequency per Operator. Then there are 64 of these multi-waveform FM-X Operators available to create a single sound – 8 operators in each of the 8 parts playable live from the keyboard. Finally each Part has powerful Filtering, DSP effects & EQ available as well. Thus there are some extremely rich synthesis capabilities lurking in Montage when using it as a modern TX816 to create a single sound with 8 FM-X Parts. So, let’s get into programming the “FM-X816” !
As to the title of this article, here I’m presenting an educational example highlighting the creation of acoustic eccentricities and “stuff” utilizing the approach of multi-Part complementary voicing. I touched on this earlier in the “Manny’s FM-Xplorations” article series, but now I’m going to go take it to another level. So, let’s pick an instrument. I decided to synthesize an acoustic instrument sound with dynamic, complex harmonic and inharmonic attack transients, a rich harmonic structure with overtone quirks that create a drastically changing timbre across the playable range, all housed within a sophisticated resonating body. Then, just for good measure throw in some significant mechanical noises from the excitation mechanism. Let’s synthesize — an acoustic Piano!I know what you’re thinking — ‘why bother’ in this age of awesome sampled and computer hosted physically modeled piano sounds. The reasons are 1) It makes for an excellent tutorial in the concepts mentioned above; 2) As a synthesized sound it has true dynamic and timbral range continuity unlike velocity switched samples; and 3) while in the end we won’t have nailed the perfect “Brand X” piano sound we will create some extremely responsive & playable piano timbres that can be simply manipulated and changed in ways that samples and physical modeling cannot.
To map out an approach to take in building our piano let’s take a general overview of it’s physical and acoustic characteristics. It’s an impulse excited string instrument, meaning energy is transferred to the strings by the act of hitting them with the keyboard hammer. There are both wound and unwound metal strings and those strings may be grouped as one, two or three per key/note depending on the piano and note range. They are strung at high tension across a metal harp, which sits inside a complex wooden resonating body. Then there is the mechanical noise from the keyboard mechanism itself, both the initial sound of the hammers hitting the strings and the key bottoming out on the keybed.
To build a piano sound “FM-X816 style” I’ll be using an approach called Harmonic Component Modeling. Harmonic Component Modeling in the context of FM synthesis means using groups of Operators to create discrete ‘building block’ parts of the overall piano sound – i.e. the general attack & body sustain; the bright & sustaining low register high order harmonics; the ‘plink’ and inharmonicities accentuated in the upper registers from the hammer attack on the short, taught strings; and key bed ‘thunk’ & noise, etc. These will all be combined together to give the completed sound. While assembling the harmonic components together, I’ll enhance the synthesis versatility of the FM-X Operators utilizing the powerful Part Insert and Master Effects and EQ in the Montage for additional timbral shaping. So, lets do it… load up the Library file “Manny FMXpert”, select the Live Set “HCM Pianos” and we’ll dive in. Also, please check out the companion video on YouTube here.
Select the Performance “MF.HCM PianoEx1”. In this example I’ll begin using two Parts to build the basic ‘body’ timbre, one Part for lower register with the wound strings and one Part for the upper register with solid strings. Additional Parts are then used for adding & enhancing the attack transients and sustaining harmonics for each register. With this example Performance I’ll be using three Parts for the lower register wound strings and one Part for the upper register solid strings to synthesize the sustaining harmonic enhancements. The remaining 2 Parts are used for the hammer attack characteristics, both the harmonic and with the inharmonic overtones along with the hammer and key bed noise across the entire range. To easily isolate and hear the various harmonic components of the total sound, I’ve set up this Performance with Scenes that turn off & on the different harmonic component Parts.
Press Assignable Switch 1 to bypass the effects then select Scene 2 to hear Parts 1 and 5 that make up the basic ‘body tone’ of the sound. Part 1 is the lower register timbre, Part 5 is for the upper range, split between keys E3 & F3. Go into Edit Mode, Select Part 1 so we can take a look. These Parts both use Algorithm 25:
We’re using Operators 3 – 8 for the basic timbre, and Operators 1 & 2 for some initial high harmonic overtones. While playing both very softly and loudly, toggle off and on Operators 1 & 2 to hear the subtle yet important character they add, then toggle off Operator 8 to here Ops 1 & 2 in isolation. I’ve done the same thing with Part 2 for the upper range, so select Part 2 and again play both soft and loud, toggle off and on Operators 1 & 2 to hear their contribution to the sound, and then toggle off Operator 8 to here Ops 1 & 2 in isolation. Ops 1 & 2 in these 2 Parts use “Stuff Trick #1” — fixed frequency mode Operators. For Ops 1 & 2 in both these Parts I’ve set the Freq Mode to Fixed, with partial pitch change per note – for reference, at a value of 99, the Fixed Mode Operator would track pitch with the notes normally as a Ratio Operator. With a value of 0 it plays the same pitch for every note, and a value of 25 would give quarter pitch note tracking. In Part 2, I’ve added a little Pitch/Vel as well. Thus, these Ops are not truly ‘fixed’ and static, but vary in a non-equal tempered manner across the keyboard and with velocity. This contributes some nice inharmonicities in the higher harmonics in the attack which is an important characteristic of acoustic sounds.
Next Select Scene 3 to isolate Part 7. This Part is adding both the harmonic and inharmonic hammer attack portion of the sound. For this Part I’m using Algorithm 74:
Each of the three Stacks is used to create a different harmonic component of the overall “hammer attack” harmonic structure. Mute Op 2, and you can hear I’ve set the Op 3-4 Stack and the four-Op Stack of Ops 5, 6, 7 & 8 create the non-harmonic hammer attack and noise portion. Mute Op 4 to hear the Op 5-8 Stack creating the noise & “thunk” portion. All these Operators are using Fixed Mode, with small amounts of Pitch/Key Scaling and Pitch/Velocity Scaling. Now, Mute Op 8 and unmute Op 4 to hear the Op 3-4 Stack ‘metallic’ attack portion — Note that as you play in the upper register above E4 you’ll start to hear the contribution of the Op 3-4 Stack creating one of the ‘acoustical eccentricities’ of a piano. Because of string length, stiffness & tuning tension, the upper register attack begins to get quite inharmonic like a percussion instrument – so I’m Fixed Frequency mode with non-integer Frequency settings and completely different amount of Pitch/Key Scaling — Op 3 set to Pitch/Key=6, and Op 4 set to Pitch/Key=39. This Op 3-4 Stack is the basis for these upper register inharmonic components, which will be refined later with Part 8. Finally, mute Ops 4 & 8, and un-mute Op 2 to hear that the Op 1-2 Stack creates the harmonic ‘hammer hitting a string’ timbre, similar to a very hard plucked string. Finally, note the Velocity sensitivity range is set to be a wide volume and harmonic velocity response to mimic the wide dynamic range in the attack of a piano. To summarize, this Part expands on our “Stuff Tricks” to synthesize more complex inharmonic components using combinations of Fixed Frequency Mode for both Carriers and multiple Modulator Operators
Let’s next select Scene 4 to isolate Part 8. This Part is adding higher order “plink” harmonics and additional key bed ‘thunk’ noise of the hammer attack across the entire key range. For this Part I’m using Algorithm 67:
Each of the 2 Op Stacks is once again creating portions of harmonic spectra components of the overall harmonic structure. Op Stacks 1-2, 3-4 and 5-6 are in Ratio Mode, using various non-integer settings for Operator frequency for the inharmonic components that play ‘pitched’ across the range – meaning they track the equal tempered pitches for each key. Ops 1 & 2 frequencies are set to 1.03 / 4.00 for the “low high” harmonics; Ops 3 & 4 frequencies are set to 13.80 / 12.72 for the “high high” harmonics; and Ops 5 & 6 frequencies are set to 9.27 / 5.00 for the “mid high” harmonics. Note how the Operator level scaling and envelope rate scalings are set so that the overall harmonic blend is sustaining and ‘buzzy’ in the low resister and transforming to a metallic ‘clank’ that quickly decays in the upper registers. As I mentioned above in looking at Part 7, this continues to expand on synthesizing the ‘acoustical eccentricities’ of the piano. Basically, because of the string length, type (wound or unwound), and stiffness & tuning tension relationships across the 88 note range, two characteristics come into play. One is the upper register behaves much more like a tuned percussion instrument than a stringed instrument – think vibraphone in the upper range vs. acoustic bass guitar (extra large!) in the lower range. For the lower registers, the non-integer tuning Ratios are to capture the ‘out of tune’ nature of the wound strings high order harmonics that skew progressively ‘sharp’, which is one of the reasons real pianos are ‘stretched tuned’. Finally Ops 7 & 9 are used for additional woody key bed ‘thunk’ noise, using Fixed Freq mode with moderate and minimal Pitch/Key tracking respectively. Thus Parts 7 and 8 combine together to synthesize these important harmonic quirks.
Finally, before moving on, select Scene 5 and play, and toggle back and forth between Scene 5 and Scene 2 to hear how using 4 of the 8 Parts — Parts 1, 2, 7 & 8 – all blending together their harmonic components begin to give us a nice basic timbre with a lot of the “stuff” and “acoustic eccentricities” of a piano. However, we’re still missing a lot of harmonic content and it’s a bit ‘thin’ and ‘digital’. Part 2 of this article series will explore how I use the remaining 4 Parts to finish building our FM-X Piano – “Fixing the Holes” in our harmonics.