--- title: "Harmonic vs Intermodulation Distortion in Audio | Gawergy Audio" description: "Compare harmonics from a single tone with sum-and-difference products from multiple tones, and learn why THD alone cannot describe a nonlinear processor on music." canonical_url: "https://gawergy.com/learn/harmonic-vs-intermodulation-distortion" md_url: "https://gawergy.com/learn/harmonic-vs-intermodulation-distortion.md" last_updated: "2026-09-23" date_published: "2026-09-23" --- # Harmonic Distortion vs Intermodulation Distortion Harmonic distortion adds components at integer multiples of an input tone. Intermodulation distortion, or IMD, arises when multiple input frequencies interact in a nonlinear system, producing new sum, difference, and higher-order combinations. A single-tone test can show the first kind while missing important aspects of the second. Music contains many simultaneous frequencies, so both mechanisms can contribute to a processor's sound. ## Key takeaways - A one-tone nonlinear test produces harmonics at multiples of the input frequency. - Two or more tones can create products related to sums and differences of their frequencies. - A THD value from one test tone cannot fully predict behavior on a dense musical signal. - Level, spectrum, measurement bandwidth, and processing design all affect the products seen and heard. ## A single tone reveals harmonic products Feed an ideal pure tone through a nonlinear system and inspect the output spectrum. In addition to the original fundamental, the output may contain energy at twice, three times, and further integer multiples of that frequency. Those are harmonics. A clipping curve provides an intuitive example: its output no longer follows a scaled copy of the sine wave, so a Fourier description needs additional frequency components. JUCE's public waveshaping tutorial demonstrates how changing a transfer function changes the sound and waveform. The [why clipping creates harmonics article](/learn/why-clipping-creates-harmonics) gives the general mechanism in more detail. The relative strengths of those components depend on the nonlinear curve, how hard it is driven, and the input. A symmetric curve applied to a centered sine may favor odd-order terms under ideal conditions; a biased or asymmetric curve may introduce even-order terms. Real music is not one centered sine, and real processors can have filters, memory, modulation, and oversampling. A harmonic series from one test tone is useful evidence, but it is not a complete portrait of how a processor will treat a drum bus or full mix. A common measurement is total harmonic distortion, or THD. It summarizes harmonic energy under a specified test. The number can be useful when test frequency, level, bandwidth, and measurement method are known. It cannot, by itself, tell you the exact spectrum of all products on complex audio, much less guarantee a listener's preference. Audio Precision's technical material notes that high-frequency harmonic measurements can even be constrained by the measurement bandwidth, motivating additional tests. ## Multiple tones expose intermodulation Now feed two tones at frequencies f1 and f2 into a nonlinear system. The output can include the original tones, their harmonics, and components such as f1+f2 and the absolute difference between them. Higher-order combinations can appear too, such as 2f1-f2 or 2f2-f1. These are intermodulation products because energy from different input components interacts. Audio Precision's amplifier-testing application note describes two-tone tests and sum-and-difference products as a way to study behavior not revealed by a single-tone THD test. A simple conceptual example uses tones at 700 Hz and 900 Hz. A nonlinear second-order term can create components at 200 Hz and 1,600 Hz, as well as harmonics of the original tones. The numbers are an illustration of arithmetic, not a suggested production setting or a prediction that every processor emits those components at the same level. Product strengths depend on the transfer function, drive, and any filtering or dynamic behavior. The [IMD deep dive](/learn/intermodulation-distortion) follows those relationships more fully. IMD can produce frequencies that are not integer multiples of either individual tone. That makes it distinct from the harmonic series of a single fundamental. Yet the distinction can blur in music because many original components already have harmonic relationships, and a combination product may coincide with a musical partial. A spectrum alone may not reveal whether a line is source content, harmonic distortion, or intermodulation without a controlled comparison. ## Why music is more complicated than either test A bass note, vocal, cymbal, and synth chord can overlap in time. When the combined waveform passes through a nonlinear curve, the curve responds to the instantaneous sum. Every component can influence how the others are reshaped. This can produce a crowded network of harmonic and intermodulation products. In a dense mix, some may be masked by existing content; others may appear as roughness, beating, or unexpected low-frequency components. The exact result depends on the audio and processor, not just the word 'distortion' on a plug-in. Changing input level can change the amount of contact with the curve. A quiet tone that barely engages a soft saturation region may interact more strongly when a loud kick is present at the same time. Thus a processor tested on isolated stems may behave differently when those stems are summed before it. This follows from nonlinear superposition failure. The [linear-versus-nonlinear processing article](/learn/linear-vs-nonlinear-audio-processing) explains why the order and summing location matter. Human perception adds another layer. An added harmonic related to one tone may sometimes reinforce pitch or timbre, while inharmonic intermodulation components may be more noticeable under some conditions. Neither category has a universal good-or-bad verdict. Audibility depends on level, masking, spectral position, arrangement, and playback. It is more honest to describe the mechanism and listen in context than to label all even harmonics pleasant and all IMD unpleasant. ## What a distortion number leaves out THD is measured with a particular stimulus and method. Two-tone IMD tests use another stimulus and can be evaluated under several standards. Audio Precision describes one high-frequency pair test and explains why its difference-frequency products can be valuable when harmonics of a high test tone leave the measurement band. A single published number without its test level, frequency, bandwidth, and reference is therefore difficult to compare fairly. Product categories, including loudspeakers and plug-ins, may use different conventions. A processor can also change with time. A compressor's detector, a dynamic saturation stage, or a modulated filter may produce behavior not captured by a static memoryless transfer curve. Sample-rate limits can fold generated harmonics or IM products into different apparent frequencies. JUCE's oversampling reference addresses the aliasing risk around nonlinear processing. An analyzer display at the project rate may therefore show products shaped by both the original nonlinearity and the digital system's frequency boundary. For a practical comparison, hold input signal, level, sample rate, and analysis method constant; compare the spectra and listen at matched output level. That is a measurement principle, not an effect recipe. A low THD reading under one tone does not promise low IMD under a complex mix. Conversely, a high measured value is not automatically objectionable in a deliberately colored instrument effect. The intended use matters. ## Harmonics and IMD are connected, not rival theories Both mechanisms arise because the output is not merely a scaled copy of the input. Mathematically, nonlinear terms that turn one sine into multiples of itself also mix different sines when they coexist. In a polynomial illustration, a squared sum contains the square of each component and a cross-product term. The cross-product is where sum-and-difference frequencies appear. This connection explains why a nonlinear curve can create both harmonic and intermodulation distortion without needing two separate mysterious circuits. The labels identify what the test reveals. One-tone harmonics refer to integer multiples of that tone. IMD refers to products involving two or more source frequencies. Neither label alone tells you whether the processing was deliberate or accidental. A clipper can be used intentionally and still produce both. A poorly overloaded converter can do so unintentionally. The [can digital clipping sound good article](/learn/can-digital-clipping-sound-good) addresses the artistic question separately from these definitions. This distinction also helps read marketing claims. 'Low THD' is meaningful for its specified conditions but should not be inflated into 'transparent on every mix.' 'Even-harmonic character' may describe a particular test, but a complex input can still generate combination products. The full system, including filtering and dynamics, determines the audible result. ## Use the right stimulus for the question Harmonic distortion describes new multiples from a tone; intermodulation distortion describes new combinations among simultaneous tones. A one-tone test makes the first easy to see, while a two-tone or multitone test reveals interaction that THD may miss. Music commonly excites both mechanisms at once. The measurements are complementary, and their interpretation depends on level, bandwidth, source material, and processor design. When a processor sounds different on a full mix than on an isolated sound, frequency interaction is one plausible cause to investigate. Measure under a stimulus relevant to the question, listen with levels matched, and resist reducing the result to a single distortion percentage or a simplistic tonal label. ## About G-Clipper Pro A clipping curve can generate both harmonics and intermodulation products on complex audio. Visual peak reduction alone does not describe the entire spectrum. ## Sources & References - [Automotive Audio Testing - Amplifiers](https://www.audioprecision.com/fileadmin-ap/technical-library/Audio-Precision-AppNote-Automotive-Audio-Amplifier-Testing.pdf) - [IMD vs. THD+N](https://www.audioprecision.com/news/imd-vs-thdn) - [Linear Time-Invariant Digital Filters — Introduction to Digital Filters](https://www.dsprelated.com/freebooks/filters/Linear_Time_Invariant_Digital_Filters.html) - [Add distortion through waveshaping and convolution](https://juce.com/tutorials/tutorial_dsp_convolution/) - [juce::dsp::Oversampling Class Template Reference](https://docs.juce.com/master/classjuce_1_1dsp_1_1Oversampling.html) - [FabFilter Saturn 2 Help — Overview](https://www.fabfilter.com/help/saturn/using/overview) - [Live Audio Effect Reference — Ableton Reference Manual Version 12](https://www.ableton.com/en/manual/live-audio-effect-reference/) ## Continue Reading - [What Is Intermodulation Distortion in Audio?](https://gawergy.com/learn/intermodulation-distortion) - [What Is Waveshaping in Audio?](https://gawergy.com/learn/waveshaping-audio) - [Why Clipping Creates Harmonics](https://gawergy.com/learn/why-clipping-creates-harmonics) ## Sitemap See the full [Gawergy.com sitemap](https://gawergy.com/sitemap.md).