FM Synthesis Research Articles

Harmonizing Innovation: A Journey through the History of Computational Music

The history of computational music traces a fascinating journey of innovation and creativity, intertwining disciplines such as computer science, mathematics, music theory, and psychology. From its inception in the mid-20th century to its current state of sophistication, computational music has evolved into a multifaceted field encompassing algorithmic composition, digital signal processing, machine learning, and interactive music systems. Pioneering works such as Roads' "The Computer Music Tutorial" and Biles' "GenJam" exemplify the early endeavors in digital sound synthesis and algorithmic composition. The roots of computational music extend back to the 1950s and 1960s with the pioneering efforts of Max Mathews, John Chowning, and Lejaren Hiller, laying the groundwork for future advancements. Technological breakthroughs in the 1970s and 1980s, including the advent of digital audio workstations and FM synthesis, further propelled the field forward. The 1990s and 2000s witnessed democratization and accessibility with the rise of affordable software tools and online collaboration platforms. Today, computational music continues to evolve with the emergence of deep learning and generative AI techniques, albeit accompanied by ethical considerations regarding copyright ownership and algorithmic bias. In conclusion, the history of computational music underscores the transformative power of technology in shaping musical expression and invites us to explore new frontiers where creativity knows no bounds.

RESEARCH OF THE PROCESS OF VISUAL ART TRANSMISSION IN MUSIC AND THE CREATION OF COLLECTIONS FOR PEOPLE WITH VISUAL IMPAIRMENTS

This article explores the creation of music through the automated generation of sounds from images. The developed automatic image sound generation method is based on the joint use of neural networks and light-music theory. Translating visual art into music using machine learning models can be used to make extensive museum collections accessible to the visually impaired by translating artworks from an inaccessible sensory modality (sight) to an accessible one (hearing). Studies of other audio-visual models have shown that previous research has focused on improving model performance with multimodal information, as well as improving the accessibility of visual information through audio presentation, so the work process consists of two parts. The result of the work of the first part of the algorithm for determining the tonality of a piece is a graphic annotation of the transformation of the graphic image into a musical series using all colour characteristics, which is transmitted to the input of the neural network. While researching sound synthesis methods, we considered and analysed the most popular ones: additive synthesis, FM synthesis, phase modulation, sampling, table-wave synthesis, linear-arithmetic synthesis, subtractive synthesis, and vector synthesis. Sampling was chosen to implement the system. This method gives the most realistic sound of instruments, which is an important characteristic. The second task of generating music from an image is performed by a recurrent neural network with a two-layer batch LSTM network with 512 hidden units in each LSTM cell, which assembles spectrograms from the input line of the image and converts it into an audio clip. Twenty-nine compositions of modern music were used to train the network. To test the network, we compiled a set of ten test images of different types (abstract images, landscapes, cities, and people) on which the original musical compositions were obtained and stored. In conclusion, it should be noted that the composition generated from abstract images is more pleasant to the ear than the generation from landscapes. In general, the overall impression of the generated compositions is positive. Keywords: recurrent neural network, light music theory, spectrogram, generation of compositions.

Disembodied Timbres: A Study on Semantically Prompted FM Synthesis

Disembodied Timbres: A Study on Semantically Prompted FM Synthesis

Abstract synthesis representation of nonlinear behaviour in coupled resonators

In this work, abstract synthesis techniques known as frequency/amplitude/phase modulation (FM, PM, and AM, respectively) are explored in musical acoustic systems involving coupled oscillators, nonlinear mechanical resonators and feedback. Though these systems are frequently modeled having acoustic/physical parameters (and units), there is a strong musical benefit of an abstract representation in which input parameters control more perceptual aspects of the produced sound. Previously coined “loopback FM" is an extension of the well known FM synthesis technique in which a carrier oscillator “loops back" to modulate its own frequency, resulting in behaviour that is characteristic of nonlinearities. With reference to loopback FM, along with its strongly related feedback PM/AM, nonlinearities that influence sounding frequency, spectral bandwidth and amplitude envelope are explored with musical instrument examples.

Bits and Pieces: A History of Chiptunes, by Kenneth B. McAlpine

<i>Bits and Pieces: A History of Chiptunes</i>, by Kenneth B. McAlpine

Creating novel tones from adjectives: An exploratory study using FM synthesis.

Creating novel tones from adjectives: An exploratory study using FM synthesis.

What Makes It Sound ’80s?

Popular music of the 1980s is remembered today as having a “sound” that is somehow unified and generalizable. The ’80s sound is tied to the electric piano preset of the Yamaha DX7 synthesizer. Not only was this preset (E. PIANO 1) astonishingly prevalent—heard in up to 61% of #1 hits on the pop, country, and R&amp;B Billboard charts in 1986—but the timbre of E. PIANO 1 also encapsulates two crucial aspects of a distinctly ’80s sound in microcosm: one, technological associations with digital FM synthesis and the Yamaha DX7 as a groundbreaking ’80s synthesizer; and two, cultural positioning in a greater lineage of popular music history. This article analyzes the timbre of E. PIANO 1 by combining ethnographic study of musician language with visual analysis of spectrograms, a novel combination of techniques that links acoustic specificity with social context. The web of connections created by the use and re-use of DX7 presets like E. PIANO 1, among hundreds or maybe thousands of different tracks and across genres, is something that allows modern listeners to abstract a unified notion of the “’80s sound” from a diverse and eclectic repertoire of songs produced in the 1980s.

Supplemental Material for Creating Novel Tones From Adjectives: An Exploratory Study Using FM Synthesis

Supplemental Material for Creating Novel Tones From Adjectives: An Exploratory Study Using FM Synthesis

Метод моделирования синтезаторов звуковых сигналов

The paper presents author’s original method of hardware synthesizers and sound processing devices modeling, taking the Yamaha TX81Z FM synthesizer as an example. The computer music language called Csound is used for the software simulation. The paper gives the basis of FM synthesis theory, the review of modeling method, and analysis of modeling results.

The Sound of Innovation: Stanford and the Computer Music Revolution by Andrew J. Nelson

Reviewed by: The Sound of Innovation: Stanford and the Computer Music Revolution by Andrew J. Nelson Simon Zagorski-Thomas (bio) The Sound of Innovation: Stanford and the Computer Music Revolution. By Andrew J. Nelson. Cambridge, MA: MIT Press, 2015. Pp. 248. $34. The Sound of Innovation provides a fascinating journey through the history of Stanford University’s Center for Computer Research in Music and Acoustics (CCRMA). Andrew Nelson has indeed drawn on “extensive archival research and dozens of interviews with digital music pioneers” (inside front cover notes) to bring this story together. One of the first keyboards I bought when I started a commercial studio in the mid-1980s was a Yamaha DX7, the hugely influential and profitable result of the partnership between Yamaha and John Chowning’s CCRMA. My first degree was a joint honors that included Artificial Intelligence and, after I’d moved away from programming toward being a “consumer” of commercial music computing, the DX7, with Chowning’s FM synthesis, was one of the few things that drew me into the more theoretical world of shaping sounds using digital processors. So, not only did I want to read this book, I also found myself drawn into the narrative and wanting it to be an epic saga of heroism. The first five chapters certainly drew me in and allowed, even encouraged, me to construct that sort of story. The remainder of the book brought me back to earth and reminded me that life is more complicated. That may not make it a very good epic saga but it does make a better history book. The CCRMA certainly deserves the sobriquet “hotbed,” and Nelson [End Page 1039] has done a great job of gathering together the evidence about what happened. I’m less convinced about the explanatory theory and, therefore, how well it fits with the MIT Inside Technology series aim to “stimulate a variety of perspectives that address the social shaping of technology.” It is not that the explanatory power of multivocality to show “how interdisciplinarity, open innovation and commercialization not only can reinforce one another, but also can form an inseparable web of mutual fate” (p. 9) is problematic. It is just that I felt there was a suggestion of agency around the CCRMA throughout the book that made that explanation more two-dimensional. So what do I mean by “a suggestion of agency”? Nelson describes multivocality as a “tactical capacity” (p. 161), a tool used by a successful agent, rather than as a generalized attribute of a system. There may be many instances of the CCRMA deliberately sustaining “multiple attributions to its activities” (p. 161) in ways that worked to its advantage. However, I wasn’t convinced that, for example, the involvement of the National Science Foundation, the Office of Naval Research, and the California Arts Council should be uncritically attributed to the “CCRMA’s skill at multivocality” (p. 162) rather than to more general attributes of the system (probably including that skill). I also felt the definition of success and progression could have been explored in more nuanced terms. The tangled web of commercial protectionism and open access can be read as an optimal combination of financial security and academic freedom in retrospect, but this doesn’t amount to the “secret formula” that Dane Stangler’s back cover quote suggests. I wonder if it amounts to much more than saying that if you want to be generous it’s a good idea to get rich first. Perhaps it’s my dour British sensibility but the narrative of superlatives grated on me. On the one hand, that may be an unpleasant by-product of me growing up without central heating in a drizzle-based climate, but it also comes from a wariness of history written by the winners. If the United Kingdom can have written its own history to the extent that the rest of the world doesn’t unanimously think of “the Empire” as a period of ruthless thievery and bloody genocide, then what are all the other victorious authors of their own history getting away with? What I should emphasize though, before I’m accused of setting up a straw man, is that I...

Enzymatic Hydrolysis of Trehalose Dimycolate Releases Free Mycolic Acids during Mycobacterial Growth in Biofilms

Mycobacterial species, like other microbes, spontaneously form multicellular drug-tolerant biofilms when grown in vitro in detergent-free liquid media. The structure of Mycobacterium tuberculosis biofilms is formed through genetically programmed pathways and is built upon a large abundance of novel extracellular free mycolic acids (FM), although the mechanism of FM synthesis remained unclear. Here we show that the FM in Mycobacterium smegmatis biofilms is produced through the enzymatic release from constitutively present mycolyl derivatives. One of the precursors for FM is newly synthesized trehalose dimycolate (TDM), which is cleaved by a novel TDM-specific serine esterase, Msmeg_1529. Disruption of Msmeg_1529 leads to undetectable hydrolytic activity, reduced levels of FM in the mutant, and retarded biofilm growth. Furthermore, enzymatic hydrolysis of TDM remains conserved in M. tuberculosis, suggesting the presence of a TDM-specific esterase in this pathogen. Overall, this study provides the first evidence for an enzymatic release of free mycolic acids from cell envelope mycolates during mycobacterial growth.

Iterative Method for Nonlinear FM Synthesis of Radar Signals

The problem of synthesizing a time-domain signal with a given energy spectral density (ESD) often arises in the field of signal processing. Many solutions have been proposed and successfully used over the years. However the problem of synthesizing a time-domain signal with constraints for a given ESD has not been investigated sufficiently. We propose a solution to one such constraint where the amplitude of the complex-valued time-domain signal is required to be unity. This is equivalent to phase modulating a unit amplitude signal, such that its ESD matches a desired energy spectral density. We provide an algorithm for this solution and apply it to a real problem encountered in radars.

Untangling Phase and Time in Monophonic Sounds

We are looking for a mathematical model of monophonic sounds with independent time and phase dimensions. With such a model we can resynthesise a sound with arbitrarily modulated frequency and progress of the timbre. We propose such a model and show that it exactly fulfils some natural properties, like a kind of time-invariance, robustness against non-harmonic frequencies, envelope preservation, and inclusion of plain resampling as a special case. The resulting algorithm is efficient and allows to process data in a streaming manner with phase and shape modulation at sample rate, what we demonstrate with an implementation in the functional language Haskell. It allows a wide range of applications, namely pitch shifting and time scaling, creative FM synthesis effects, compression of monophonic sounds, generating loops for sampled sounds, synthesise sounds similar to wavetable synthesis, or making ultrasound audible.

A quadrature baseband approach to direct digital FM synthesis

The use of direct digital synthesis (DDS) to synthesize high-frequency analog signals has received much attention over the past decade. This technique allows modulation schemes such as FM and PM to be implemented with high fidelity using digital components. This paper investigates an improvement on the typical DDS approach of synthesizing the desired signal at an intermediate frequency. Instead, two quadrature baseband signals are generated, which are then mixed directly to the output center frequency. This method can be used to synthesize FM signals of arbitrarily high output frequency, using digital circuitry with considerably lower clock speeds.

Chaotic Sound Synthesis

Solutions of chaotic and fractal equations have provided artistically new and interesting classes of images and sounds. Musical scores have been produced from Mandelbrot and Julia set images, and chaotic and fractal techniques have been used to directly synthesize a variety of sound waveforms (see, for example, Monro 1991). One popular approach to sound synthesis is based on Chua's circuit (Mayer-Kress et al. 1993; Mayer-Kress, Choi, and Bargar 1993; Hunt and Johnson 1993). Another method, described here, uses cross-coupled frequency-modulated oscillators (chaotic FM synthesis). Sounds produced by the chaotic FM technique vary from pure sine waves to a variety of complex signals and colored noises. The chaotic FM algorithm includes conventional FM synthesis as a subset. This chaotic FM method can be combined with other chaotic synthesis methods, such as the Ueda attractor, providing an even wider range of sounds. Chaotic synthesis and filtering can be used in both the analog and digital domains, providing similar results. Several examples of chaotic synthesis methods in both domains are provided in this article. The emphasis, however, is on the analog methods, with examples using the Moog modular synthesizer, Buchla electronic musical instruments, and electronic analog computers.

Spectral modelling and timbre hybridisation programs for computer music

Software packages developed for visualising time-varying timbres, finding parameter settings for modelling acoustic instrument tones, and for synthesising timbres are described. The sndan package, written in C for Unix machines, provides spectrum analysis, pitch tracking, spectrum display, parameter modification, and additive synthesis. Other programs that estimate the best possible parameters for recreating acoustic sounds using techniques such as multiple FM and wavetable synthesis and extended nonlinear/filter synthesis are outlined. Hybrid instruments created by fashioning sounds out of the recombined spectral characteristics of two or more very different tones are also presented.

Sonic Foundry Sound Forge Version 3.0 Sample Editing Software for Windows

For years, DOS-based sampling editing software has been the only DOS music or audio program on my computer. I had long since made the transition to Windows versions of everything else, but I had been unable to find a suitable Windows sample editor. No longer. Sonic Foundry's Sound Forge Version 3.0 provides a complete set of sample editing, transfer, and conversion tools behind an easyto-use Windows shell. The program works with any PC-compatible sound card, and supports a variety of widely used samplers, as well as generic MIDI sample dump standard. Sound Forge uses the same visual representation as other programs of its kind, displaying audio waveform data in a window as either a single mono track or a stereo pair, with the left-channel data displayed above the right channel. The user typically selects a particular region of one or both tracks before performing any of a variety of operations on the selected audio data. Those operations include cut, copy, and paste, as well as numerous other editing and processing functions, including mix and cross-fade, time compress/expand, fade in/out, and equalization. The user can mark individual points within a sound file, or identify particular regions within a file and perform operations on a specified region. Sound Forge also supports playlist editing, though only within a single sound file. The range and variety of processing functions is at the core of Sound Forge's strength. It emulates a studio-full of signal processing gear, and, of course, does all its processing in the digital domain. The effects include amplitude modulation, chorus, delay/echo, distortion, noise gate, pitch bend, pitch change, reverberation, and vibrato, as well as such tools as graphic and parametric equalization, and four-operator FM synthesis. Each of the functions and tools incorporates an impressive amount of user control over parameter values, though many of the tools also offer default settings or predefined parameter options. There are so many tools and so many options in each, that in two months of intensive use I was not able to explore every one. Typically, the trade-off between hardware and software signal processing is speed, and Sound Forge is no different. On short samples, processing time for many operations is quite fast, though processing speed is determined both by the complexity of the function and the capability of the host computer. Running on a 80486 DX/2-66, a linear fade out over a 4-sec sample took approximately 2 sec. Adding reverberation to a 5-sec sample took slightly more than 12 sec. I used Sound Forge to edit and mix a long-form audio project, and clearly pushed the program to (and occasionally past) its limits. One of the limiting factors is hard-disk use. Sound Forge creates disk backups of each open sound-file window, as well as undo information and clipboard data. A single mix, for example, might require that several windows be open at once, so a large, complex mix can make tremendous demands on hard-disk resources. To its credit, Sound Forge does not set an arbitrary maximum number of sample-data windows that may be open at once, so the limit on the number of open windows is a function of the hardware system limitations. As I worked with larger files (up to 10 min long), processing times became correspondingly longer. Compressing or equalizing the file could take 8-10 min-not an inter-

Progress toward more intuitive programming of FM synthesizers

As was reported last year, work has been done to make the programming of new tone colors on FM synthesizers more intuitive for composers and musicians. The logarithmic relations between modulator, carrier, and envelope level parameters on commercially available instruments and corresponding index of modulation and output amplitudes in standard FM synthesis equations have been empirically determined. The relations have been incorporated into a tutorial computer program that displays the waveform and corresponding spectrum after synthesis parameters are entered in either form. With the program, users can gain familiarity with the spectra of FM‐synthesized sounds, which can be quite complicated for synthesizers using four or six oscillators in various configurations. Work is currently being done to calibrate feedback and envelope rate parameters and to develop software to go in the opposite direction, determining FM parameters to approximate spectra with complicated and time‐varying shapes. Ultimately, such software should choose the configuration of oscillators as well as other parameters to create the perceptually closest approximation to the desired spectral characteristics.

FM synthesis in musical instruments

The objective of the presentation will be to explain the development and implementation of FM synthesis in musical instruments. Topics to be covered include: (1) development of FM synthesis; (2) introduction of FM synthesis in musical instruments; (3) explanation and demonstration of principles of FM synthesis; (4) implementation of FM synthesis in current Yamaha instruments: the DX7; (5) demonstration of the musical capabilities of FM synthesis.

Interferometric Direction Finding on an FM/CW Ionosonde

DF interferometry on an FM/CW sounder has been successfully performed using a three‐element crossed baseline array and twin channel receiver modified for linear FM synthesis. Measured baseline phase differences obtained from a sum‐and‐difference CRT phase display were used to compute incident azimuth and elevation. Phase ambiguities were resolved by prior knowledge of the sounder location. Measurements in San Antonio, Texas, on cooperative FM/CW targets in Pueblo and Boulder, Colorado, showed essentially constant azimuth within ±2°, commensurate with ionospheric lateral deviation effects. Corresponding elevation data showed increasing elevation angle with frequency indicative of increasing ray penetration. The azimuth/elevation measurements were made without benefit of IF spectral analysis for time‐delay‐mode resolution. Consequently, measurements were successful over those frequency ranges at which the sounder signal dominated interfering stations. This limitation would be eliminated by the use of cross‐spectrum analysis at the receiver output.