{ "id": "ep6t5sb671", "slug": "time-frequency-representation", "title": "Time–frequency representation", "updated": "2026-09-30", "topic_path": [ { "id": "technology", "label": "Technology and the built world", "api_url": "https://www.edgepedia.org/api/v1/topics/technology" }, { "id": "technology.computing", "label": "Computing and digital systems", "api_url": "https://www.edgepedia.org/api/v1/topics/technology.computing" }, { "id": "technology.computing.ai", "label": "Artificial intelligence and data", "api_url": "https://www.edgepedia.org/api/v1/topics/technology.computing.ai" }, { "id": "technology.computing.ai.algorithms", "label": "Algorithms and computational methods", "api_url": "https://www.edgepedia.org/api/v1/topics/technology.computing.ai.algorithms" }, { "id": "technology.computing.ai.numerical_string_geometry", "label": "Numerical, string, and geometric algorithms", "api_url": "https://www.edgepedia.org/api/v1/topics/technology.computing.ai.numerical_string_geometry" }, { "id": "technology.computing.ai.numerical_string_geometry.fourier_signal_transforms", "label": "Fourier and signal transforms", "api_url": "https://www.edgepedia.org/api/v1/topics/technology.computing.ai.numerical_string_geometry.fourier_signal_transforms" } ], "geo": [ { "id": "geo.nongeo.t1946.technology.computing.ai.numerical_string_geometry.fourier_signal_transforms", "label": "Non-geographic · 1946 to 2000: Fourier and signal transforms", "api_url": "https://www.edgepedia.org/api/v1/geo/geo.nongeo.t1946.technology.computing.ai.numerical_string_geometry.fourier_signal_transforms", "path": [ { "id": "geo.nongeo", "label": "Non-geographic", "api_url": "https://www.edgepedia.org/api/v1/geo/geo.nongeo" }, { "id": "geo.nongeo.t1946", "label": "Non-geographic · 1946 to 2000", "api_url": "https://www.edgepedia.org/api/v1/geo/geo.nongeo.t1946" }, { "id": "geo.nongeo.t1946.technology", "label": "Technology and the built world", "api_url": "https://www.edgepedia.org/api/v1/geo/geo.nongeo.t1946.technology" }, { "id": "geo.nongeo.t1946.technology.computing", "label": "Computing and digital systems", "api_url": "https://www.edgepedia.org/api/v1/geo/geo.nongeo.t1946.technology.computing" }, { "id": "geo.nongeo.t1946.technology.computing.ai", "label": "Artificial intelligence and data", "api_url": "https://www.edgepedia.org/api/v1/geo/geo.nongeo.t1946.technology.computing.ai" }, { "id": "geo.nongeo.t1946.technology.computing.ai.algorithms", "label": "Algorithms and computational methods", "api_url": "https://www.edgepedia.org/api/v1/geo/geo.nongeo.t1946.technology.computing.ai.algorithms" }, { "id": "geo.nongeo.t1946.technology.computing.ai.numerical_string_geometry", "label": "Numerical, string, and geometric algorithms", "api_url": "https://www.edgepedia.org/api/v1/geo/geo.nongeo.t1946.technology.computing.ai.numerical_string_geometry" }, { "id": "geo.nongeo.t1946.technology.computing.ai.numerical_string_geometry.fourier_signal_transforms", "label": "Fourier and signal transforms", "api_url": "https://www.edgepedia.org/api/v1/geo/geo.nongeo.t1946.technology.computing.ai.numerical_string_geometry.fourier_signal_transforms" } ] } ], "excerpt": "A time–frequency representation is a signal processing method that maps a signal's energy over time and frequency at once, with the spectrogram as the standard example.", "snippet": "A time–frequency representation is a signal processing method that maps a signal's energy over time and frequency at once, with the spectrogram as the standard example.", "node": "technology.computing.ai.numerical_string_geometry.fourier_signal_transforms", "markdown": "# Time–frequency representation\n\nA time–frequency representation (TFR) is a signal processing method that maps a signal's energy simultaneously over time and frequency, showing how its frequency content changes from moment to moment. The spectrogram has long been the standard method for the analysis of non-stationary signals such as speech, radar returns, and brain recordings.[1](https://dspace.mit.edu/bitstream/handle/1721.1/50243/Kashif_Techniques%20to%20Obtain.pdf?sequence=1&isAllowed=y) The choice of representation is not unique: no single TFR satisfies all the properties of a physically correct joint time–frequency energy density, and the choice cannot be made on mathematical analysis alone.[2](https://apps.dtic.mil/sti/tr/pdf/ADA385576.pdf)\n\n| Key fact | Detail |\n|---|---|\n| Output | A two-dimensional energy map over time and frequency (\"information diagrams\")[3](https://digital-library.theiet.org/content/journals/10.1049/ji-3-2.1946.0074) |\n| Uncertainty limit | A time–frequency atom's box has area at least 1/2, reached only by a Gaussian window; other conventions write \\( B \\cdot T \\geq 1/(4\\pi) \\) or \\( T \\cdot B = 1 \\)[4](https://www.di.ens.fr/~mallat/College/Chap4-Time-Frequency.pdf) • [5](https://ar5iv.labs.arxiv.org/html/2101.06707) • [6](https://tftb.nongnu.org/tutorial.pdf) |\n| Spectrogram | Squared magnitude of the windowed Fourier transform; the STFT is exactly invertible when the analysis window and hop size satisfy a constant-overlap-add constraint[7](https://www.maths.lu.se/fileadmin/maths/matematisk_statistik/FMSF10/lecture6_2020.pdf) • [6](https://tftb.nongnu.org/tutorial.pdf) • [34](https://ccrma.stanford.edu/~jos/sasp/FBS_Perfect_Reconstruction.html) |\n| Wigner–Ville distribution | Twice the concentration of the optimal spectrogram, but cross terms up to twice the auto-term amplitude[7](https://www.maths.lu.se/fileadmin/maths/matematisk_statistik/FMSF10/lecture6_2020.pdf) • [5](https://ar5iv.labs.arxiv.org/html/2101.06707) |\n| Cohen's class | All quadratic TFDs obtained by ambiguity-domain kernels; the WVD corresponds to kernel φ(ν, τ) = 1[7](https://www.maths.lu.se/fileadmin/maths/matematisk_statistik/FMSF10/lecture6_2020.pdf) |\n| Synchrosqueezing | Reassigns only in frequency, is reversible, and supports mode reconstruction[8](https://pmc.ncbi.nlm.nih.gov/articles/PMC6048578/) |\n| Main users | Speech, radar and sonar, biomedical signals, machinery vibration, seismology, power systems[9](https://tfsa.ucg.ac.me/pap/tfsa-001052.pdf) • [10](https://www.sciencedirect.com/science/article/abs/pii/S1051200414003571) |\n\n## How it works\n\nA TFR estimates how the signal's energy is distributed over the time–frequency plane. Every method is bounded by the uncertainty principle: the Heisenberg uncertainty theorem shows the area of a time–frequency atom's box is at least 1/2, and only a Gaussian window reaches that minimum.[4](https://www.di.ens.fr/~mallat/College/Chap4-Time-Frequency.pdf) Published sources state the same constraint with different numerical conventions: \\( B \\cdot T \\geq 1/(4\\pi) \\), with B the bandwidth of one frequency bin,[5](https://ar5iv.labs.arxiv.org/html/2101.06707) and \\( T \\cdot B = 1 \\) for Gaussian functions.[6](https://tftb.nongnu.org/tutorial.pdf) The consequence is method-independent: a signal cannot have an arbitrarily small support in time and in frequency at once, so good time resolution requires a short window and good frequency resolution a long one.[6](https://tftb.nongnu.org/tutorial.pdf) The spectrogram, the most used TFR, is therefore a biased estimator of instantaneous frequency and group delay.[1](https://dspace.mit.edu/bitstream/handle/1721.1/50243/Kashif_Techniques%20to%20Obtain.pdf?sequence=1&isAllowed=y)\n\n## How it is done\n\nThe short-time [Fourier transform](https://www.edgepedia.org/fourier-transform) pre-windows the signal around each time instant, computes the Fourier transform, and repeats for every instant:[6](https://tftb.nongnu.org/tutorial.pdf)\n\n\\[ X(\\tau, f) = \\int_{-\\infty}^{\\infty} x(t)\\, w(t-\\tau)\\, e^{-i 2\\pi f t}\\, dt \\]\n\nThe spectrogram is its squared magnitude, \\( S_{x}(t, f) = |X(t, f)|^{2} \\), with a unit-energy window h(t) centered at time t.[7](https://www.maths.lu.se/fileadmin/maths/matematisk_statistik/FMSF10/lecture6_2020.pdf) Window length is the main design choice: a rule of thumb sets it to the component length, since short windows smear components in frequency and long windows smear low-frequency components in time; the FFT length L should be a power of 2 larger than the window length M, with step \\( N_{\\mathrm{step}} \\) often chosen as M/8.[11](https://www.maths.lu.se/fileadmin/maths/personal_staff/mariasandsten/TFkompver4.pdf) Window shape controls leakage: rectangular windows have the narrowest main lobe and largest side lobes, Hann is moderate, and Blackman has the widest main lobe with the strongest sidelobe suppression.[12](https://compneuro.neuromatch.io/tutorials/W2D2_SignalProcessing/student/W2D2_Tutorial4.html) For chirps, the optimal window width is inversely proportional to the instantaneous chirp rate, \\( \\sigma \\propto 1/\\text{chirp rate} \\).[13](https://link.springer.com/article/10.1186/s13634-025-01287-8)\n\n## Origin\n\nThe windowed Fourier transform with a Gaussian window, and the \"elementary signals\" (logons) occupying the smallest possible area \\( \\Delta t \\cdot \\Delta f = 1/2 \\), each conveying one \"quantum of information\", were presented by [Dennis Gabor](https://www.edgepedia.org/dennis-gabor) in his 1946 paper \"Theory of communication. Part 1: The analysis of information\" in the Journal of the Institution of Electrical Engineers.[3](https://digital-library.theiet.org/content/journals/10.1049/ji-3-2.1946.0074) Gabor's original parameter choice \\( \\omega_{0} \\cdot t_{0} = 2\\pi \\) leads to unstable reconstruction.[14](https://sites.math.duke.edu/~ingrid/publications/ieee36-1990.pdf) The Wigner–Ville distribution, the central member of Cohen's class (the set of bilinear representations covariant under time–frequency translations), uses the analytic signal.[15](https://www.iste.co.uk/data/doc_kmwlrnocsxjk.pdf) Reassignment was pioneered by Kunihiko Kodera, Claude De Villedary, and Roger Gendrin in a 1976 paper in Physics of The Earth and Planetary Interiors under the name Modified Moving Window Method;[16](https://doi.org/10.1016/0031-9201%2876%2990044-3) • [17](https://ar5iv.labs.arxiv.org/html/0903.3080) F. Auger and P. Flandrin coined the term \"reassignment\" and generalized the method to Cohen's class and time–scale representations in 1995 in IEEE Transactions on Signal Processing.[18](https://doi.org/10.1109/78.382394) The S-transform was reported by R.G. Stockwell, L. Mansinha, and R.P. Lowe in 1996 in IEEE Transactions on Signal Processing.[19](https://doi.org/10.1109/78.492555) Synchrosqueezing was reported by [Ingrid Daubechies](https://www.edgepedia.org/ingrid-daubechies) and Stéphane Maes in an auditory-receptive context,[20](https://doi.org/10.1201/9780203734032-20) and the synchrosqueezed wavelet transform as an empirical-mode-decomposition-like tool with a mode-reconstruction theorem by Ingrid Daubechies, Jianfeng Lu, and Hau-Tieng Wu in 2010 in Applied and Computational Harmonic Analysis.[21](https://doi.org/10.1016/j.acha.2010.08.002) Empirical mode decomposition was proposed by [Norden E. Huang](https://www.edgepedia.org/norden-e-huang) and colleagues in 1998 in Proceedings of the Royal Society A.[22](https://doi.org/10.1098/rspa.1998.0193)\n\n## Variants\n\n**Spectrogram.** Quadratic, and thus not linear or additive in the signal, with possible cross terms between components, but limited by the uncertainty principle to modest joint resolution.[23](https://dl.acm.org/doi/10.1016/j.dsp.2015.03.004) • [35](https://pmc.ncbi.nlm.nih.gov/articles/PMC10346397/)\n\n**Wigner–Ville distribution.** Defined as \\( W_{x}(t, f) = \\int_{-\\infty}^{\\infty} x(t + \\tau/2)\\, x^{*}(t - \\tau/2)\\, e^{-i 2\\pi f \\tau}\\, d\\tau \\), usually computed on the analytic signal \\( z(t) = s(t) + jH\\{s(t)\\} \\) to avoid cross terms between positive and negative frequencies.[7](https://www.maths.lu.se/fileadmin/maths/matematisk_statistik/FMSF10/lecture6_2020.pdf) • [10](https://www.sciencedirect.com/science/article/abs/pii/S1051200414003571) For a Gaussian signal, the Wigner distribution has twice the concentration of a spectrogram using the optimal matching Gaussian window, that is half the time–frequency spread.[7](https://www.maths.lu.se/fileadmin/maths/matematisk_statistik/FMSF10/lecture6_2020.pdf) • [11](https://www.maths.lu.se/fileadmin/maths/personal_staff/mariasandsten/TFkompver4.pdf) The price is interference: by Janssen's formula, the interference term of two components localized at \\( (t_{1}, f_{1}) \\) and \\( (t_{2}, f_{2}) \\) is localized at the center point \\( ((t_{1}+t_{2})/2, (f_{1}+f_{2})/2) \\), and cross terms may have twice the amplitude of the auto terms with an oscillatory pattern.[15](https://www.iste.co.uk/data/doc_kmwlrnocsxjk.pdf) • [5](https://ar5iv.labs.arxiv.org/html/2101.06707)\n\n**Cohen's class.** Members are generated from the ambiguity function through a kernel \\( \\varphi(\\nu, \\tau) \\); the WVD is the kernel \\( \\varphi(\\nu, \\tau) = 1 \\), and preserving the marginals requires \\( \\varphi(0, \\tau) = \\varphi(\\nu, 0) = 1 \\).[7](https://www.maths.lu.se/fileadmin/maths/matematisk_statistik/FMSF10/lecture6_2020.pdf) The Choi–Williams (exponential) distribution uses \\( \\varphi(\\nu, \\tau) = e^{-\\nu^{2} \\cdot \\tau^{2}/\\sigma} \\); the smoothed pseudo-WVD filters with separate kernels g(t) and H(f), and Choi–Williams, Margenau–Hill, and Rihaczek variants often give very similar practical results to the SPWVD.[7](https://www.maths.lu.se/fileadmin/maths/matematisk_statistik/FMSF10/lecture6_2020.pdf) • [5](https://ar5iv.labs.arxiv.org/html/2101.06707) The spectrogram itself is a smoothed WVD whose kernel is the Wigner–Ville distribution of the window.[17](https://ar5iv.labs.arxiv.org/html/0903.3080) A published comparison table recommends STFT/Gabor for general-purpose use, the CWT when variable resolution is required, the [Stockwell transform](https://www.edgepedia.org/stockwell-transform) when fixed phase alignment is needed, the SPWVD when high resolution with tolerable artifacts is needed, and the WVD only for simple signals or when artifacts can be tolerated.[5](https://ar5iv.labs.arxiv.org/html/2101.06707)\n\n**Wavelet and related transforms.** The continuous wavelet transform uses short windows at high frequencies and long windows at low frequencies (constant Q = B/ν); its scalogram has frequency resolution finer than the spectrogram at low frequencies but coarser at higher frequencies.[6](https://tftb.nongnu.org/tutorial.pdf) • [4](https://www.di.ens.fr/~mallat/College/Chap4-Time-Frequency.pdf) The S-transform is an STFT with a frequency-dependent Gaussian window, carries absolute referenced phase, and tends to emphasize higher frequencies through the factor |f| in its formula.[5](https://ar5iv.labs.arxiv.org/html/2101.06707) The superlet transform combines multiple wavelet representations by geometric average to achieve time–frequency super-resolution.[24](https://www.frontiersin.org/articles/10.3389/fninf.2022.871904/pdf)\n\n**Reassignment and synchrosqueezing.** Reassignment refocuses blurred spectrogram energy toward the true region of support of the signal.[17](https://ar5iv.labs.arxiv.org/html/0903.3080) Synchrosqueezing transfers information from the time–scale plane to the time–frequency plane via the map (b, a) → (b, ωₛ(a, b)), estimating instantaneous frequency as \\( \\omega_{s}(a, b) = -i\\,[W_{s}(a, b)]^{-1}\\, \\partial W_{s}(a, b)/\\partial b \\); it is a special case of reassignment that reassigns only in frequency and additionally enables mode reconstruction.[25](https://www.math.ucdavis.edu/~saito/data/synchrosqueezing/daubechies-lu-wu_synchrosqueezing.pdf) • [26](https://hal.science/hal-00983755v1/document) Linear TFRs are additive (the TFR of a sum equals the sum of the TFRs) and allow component extraction and reconstruction, which is problematic for quadratic representations.[23](https://dl.acm.org/doi/10.1016/j.dsp.2015.03.004)\n\n## Applications\n\nDemonstrated uses of the reassigned spectrogram include speech phonation analysis, whale song pitch tracking, and additive sound modeling.[27](https://pubs.aip.org/asa/jasa/article/119/1/360/538332/Algorithms-for-computing-the-time-corrected) [Reference](https://www.edgepedia.org/reference) texts list radar and sonar processing, biomedicine, multimedia, telecommunications, seismology, car engine technology, and optics.[9](https://tfsa.ucg.ac.me/pap/tfsa-001052.pdf) In biomedicine, TFD-based abnormality detection of physiological signals (PCG, ECG, EEG, HRV) outperforms time-domain or frequency-domain-only approaches because TFDs match the non-stationary character of these signals.[10](https://www.sciencedirect.com/science/article/abs/pii/S1051200414003571) Synchrosqueezed transforms have been applied in geophysics, paleoclimatic studies, medical studies, mechanical engineering, and financial studies, with demonstrations on the LIGO gravitational-wave signal and volcano-seismic tremor from [Popocatépetl](https://www.edgepedia.org/popocatepetl) volcano.[8](https://pmc.ncbi.nlm.nih.gov/articles/PMC6048578/) • [26](https://hal.science/hal-00983755v1/document) HF radar data have been analyzed by comparing the STFT, S-transform, and Wigner distribution.[2](https://apps.dtic.mil/sti/tr/pdf/ADA385576.pdf) Deep-learning models now synthesize TFRs directly: QTFN, an end-to-end quadratic time–frequency network by Tao Chen and colleagues (2024, Big Data Mining and [Analytics](https://www.edgepedia.org/analytics)), generates data-driven basis functions and targets cross-term-free, high-resolution quadratic TFDs, trained only on synthetic signals yet tested on synthetic and real-world data.[28](https://www.sciopen.com/article/10.26599/BDMA.2024.9020031) On the adaptive side, MATFWSET by Jen-Chieh Cheng and Jian-Jiun Ding (2026, Journal on Advances in Signal Processing) proposes a reliable window-width range and a multitaper combination of time-varying and frequency-varying window widths, choosing the optimal width for each time–frequency point.[13](https://link.springer.com/article/10.1186/s13634-025-01287-8)\n\n## Limitations and alternatives\n\nThe Wigner–Ville distribution is highly concentrated but highly nonlinear and non-local, very sensitive to noise, and generates cross-components that often mask the components of interest in multicomponent signals.[17](https://ar5iv.labs.arxiv.org/html/0903.3080) All quadratic TFDs suffer an inherent compromise between cross-term suppression and auto-term resolution, which can degrade feature extraction from time–frequency images.[10](https://www.sciencedirect.com/science/article/abs/pii/S1051200414003571) Reassignment does not increase resolving power: components smeared together by the analysis window remain smeared, and the reassigned representation is no longer invertible, so multicomponent modes cannot easily be retrieved.[17](https://ar5iv.labs.arxiv.org/html/0903.3080) • [29](https://comptes-rendus.academie-sciences.fr/physique/item/10.1016/j.crhy.2019.07.001.pdf) Synchrosqueezing works well only when the modes are slightly modulated, a major limitation for radar, speech, gravitational waves, and otoacoustic emissions, and it is not suited to signals with continuous broad-band spectra; the second-order synchrosqueezing transform (FSST2) was introduced to handle strongly frequency-modulated modes.[29](https://comptes-rendus.academie-sciences.fr/physique/item/10.1016/j.crhy.2019.07.001.pdf) • [8](https://pmc.ncbi.nlm.nih.gov/articles/PMC6048578/) Post-processing methods are constrained by their basis (STFT or WT): when two frequency components are closely adjacent or overlapped, the results have poor resolution regardless of post-processing.[30](https://www.sciencedirect.com/science/article/abs/pii/S0165168423003213) A detailed numerical study found that the higher concentration of synchrosqueezed transforms does not imply better resolution, and that even noise-free synchrosqueezed TFRs contain many small-amplitude spurious spikes, so component amplitude cannot be read from peak heights.[23](https://dl.acm.org/doi/10.1016/j.dsp.2015.03.004)\n\n**Alternatives.** [Empirical mode decomposition](https://www.edgepedia.org/empirical-mode-decomposition) coupled with the [Hilbert transform](https://www.edgepedia.org/hilbert-transform) (the Hilbert–Huang spectrum) achieves higher time–frequency localization than STFT/CWT, but has mode mixing and splitting, aliasing, and end-point artifacts, lacks mathematical foundations, and behaves like a filter bank.[31](https://imgw.univie.ac.at/fileadmin/user_upload/i_img/Geophyik/Publikationen_bis_2018/Tary_2014_Spectral.pdf) • [29](https://comptes-rendus.academie-sciences.fr/physique/item/10.1016/j.crhy.2019.07.001.pdf) [Matching pursuit](https://www.edgepedia.org/matching-pursuit) decomposes a signal into waveforms selected from a dictionary of time–frequency atoms, and its energy distribution contains no interference terms; in one comparison it produced an extremely poor TFR for a mono-component signal with sinusoidal instantaneous-frequency law, while the Choi–Williams distribution followed the law with high resolution and minimal cross terms.[1](https://dspace.mit.edu/bitstream/handle/1721.1/50243/Kashif_Techniques%20to%20Obtain.pdf?sequence=1&isAllowed=y) • [32](https://eurasip.org/Proceedings/Eusipco/Eusipco2005/defevent/papers/cr1246.pdf) The spectrogram performs poorly for newborn EEG seizure signals because of its poor resolution, and MP-based signal-adaptive approaches are preferred in biomedical work because they provide appropriate resolution at all frequencies while reducing cross terms.[32](https://eurasip.org/Proceedings/Eusipco/Eusipco2005/defevent/papers/cr1246.pdf) • [33](https://pubmed.ncbi.nlm.nih.gov/23703538/) A quantitative comparison found adaptive quadratic TFRs give the best overall performance when no a priori information about the signal is known.[32](https://eurasip.org/Proceedings/Eusipco/Eusipco2005/defevent/papers/cr1246.pdf) Jones and Parks concluded that no TFD is best for all time–frequency analysis and that concentration and resolution cannot both be improved at once.[1](https://dspace.mit.edu/bitstream/handle/1721.1/50243/Kashif_Techniques%20to%20Obtain.pdf?sequence=1&isAllowed=y)\n\n## References\n\n1. [Techniques to Obtain Good Resolution and Concentration in Time-Frequency Distributions (review chapter, MIT DSpace)](https://dspace.mit.edu/bitstream/handle/1721.1/50243/Kashif_Techniques%20to%20Obtain.pdf?sequence=1&isAllowed=y)\n2. [Linear and Quadratic Time-Frequency Representations (DTIC technical report)](https://apps.dtic.mil/sti/tr/pdf/ADA385576.pdf)\n3. [Theory of communication. Part 1: The analysis of information (Gabor, 1946)](https://digital-library.theiet.org/content/journals/10.1049/ji-3-2.1946.0074)\n4. [A Wavelet Tour of Signal Processing, Chapter 4: Time-Frequency (Mallat)](https://www.di.ens.fr/~mallat/College/Chap4-Time-Frequency.pdf)\n5. [Fourier, Gabor, Morlet or Wigner: Comparison of Time-Frequency Transforms](https://ar5iv.labs.arxiv.org/html/2101.06707)\n6. [Time-Frequency Toolbox Tutorial (TFTB)](https://tftb.nongnu.org/tutorial.pdf)\n7. [Introduction to time-frequency analysis (Lund University lecture notes, FMSF10)](https://www.maths.lu.se/fileadmin/maths/matematisk_statistik/FMSF10/lecture6_2020.pdf)\n8. [Analysis of time-varying signals using continuous wavelet and synchrosqueezed transforms (PMC)](https://pmc.ncbi.nlm.nih.gov/articles/PMC6048578/)\n9. [Time-Frequency Signal Analysis (Stanković et al., book front matter and TOC)](https://tfsa.ucg.ac.me/pap/tfsa-001052.pdf)\n10. [Time–frequency features for pattern recognition using high-resolution TFDs: A tutorial review (Digital Signal Processing, Elsevier)](https://www.sciencedirect.com/science/article/abs/pii/S1051200414003571)\n11. [Time-Frequency Analysis (Maria Sandsten, Lund University course compendium)](https://www.maths.lu.se/fileadmin/maths/personal_staff/mariasandsten/TFkompver4.pdf)\n12. [Tutorial 4: Time-Frequency Methods, Neuromatch Academy](https://compneuro.neuromatch.io/tutorials/W2D2_SignalProcessing/student/W2D2_Tutorial4.html)\n13. [Multitaper adaptive time–frequency windowed synchroextracting transform (EURASIP JASP, 2025)](https://link.springer.com/article/10.1186/s13634-025-01287-8)\n14. [The wavelet transform, time-frequency localization and signal analysis (Daubechies, IEEE Trans. Inf. Theory, 1990)](https://sites.math.duke.edu/~ingrid/publications/ieee36-1990.pdf)\n15. [Quadratic Time-Frequency Analysis I: Cohen's class and the Wigner-Ville distribution (ISTE book chapter)](https://www.iste.co.uk/data/doc_kmwlrnocsxjk.pdf)\n16. [A new method for the numerical analysis of non-stationary signals (Physics of The Earth and Planetary Interiors, 1976)](https://doi.org/10.1016/0031-9201%2876%2990044-3)\n17. [A Unified Theory of Time-Frequency Reassignment (Fulop & Fitz)](https://ar5iv.labs.arxiv.org/html/0903.3080)\n18. [F. Auger, P. Flandrin (1995). Improving the readability of time-frequency and time-scale representations by the reassignment method. IEEE Transactions on Signal Processing.](https://doi.org/10.1109/78.382394)\n19. [R.G. Stockwell, L. Mansinha, R.P. Lowe (1996). Localization of the complex spectrum: the S transform. IEEE Transactions on Signal Processing.](https://doi.org/10.1109/78.492555)\n20. [Ingrid Daubechies, Stéphane Maes (2017). A Nonlinear Squeezing of the Continuous Wavelet Transform Based on Auditory Nerve Models. .](https://doi.org/10.1201/9780203734032-20)\n21. [Ingrid Daubechies, Jianfeng Lu, Hau-Tieng Wu (2010). Synchrosqueezed wavelet transforms: An empirical mode decomposition-like tool. Applied and Computational Harmonic Analysis.](https://doi.org/10.1016/j.acha.2010.08.002)\n22. [Norden E. Huang and colleagues (1998). The empirical mode decomposition and the Hilbert spectrum for nonlinear and non-stationary time series analysis. Proceedings of the Royal Society A Mathematical Physical and Engineering Sciences.](https://doi.org/10.1098/rspa.1998.0193)\n23. [Linear and synchrosqueezed time-frequency representations revisited (Digital Signal Processing, 2016)](https://dl.acm.org/doi/10.1016/j.dsp.2015.03.004)\n24. [Time-Frequency Representations of Brain Oscillations: Which One Is Better? (Frontiers in Neuroscience, 2022)](https://www.frontiersin.org/articles/10.3389/fninf.2022.871904/pdf)\n25. [Synchrosqueezed wavelet transforms: An empirical mode decomposition-like tool (Daubechies, Lu, Wu, ACHA 2011)](https://www.math.ucdavis.edu/~saito/data/synchrosqueezing/daubechies-lu-wu_synchrosqueezing.pdf)\n26. [Time-Frequency Reassignment and Synchrosqueezing: An Overview (Auger, Flandrin, Lin, McLaughlin, Meignen, Oberlin, Wu; IEEE SPM 2013)](https://hal.science/hal-00983755v1/document)\n27. [Algorithms for computing the time-corrected instantaneous frequency (reassigned) spectrogram, with applications (Fulop & Fitz, JASA 2006)](https://pubs.aip.org/asa/jasa/article/119/1/360/538332/Algorithms-for-computing-the-time-corrected)\n28. [QTFN: A General End-to-End Time-Frequency Network (Big Data Mining and Analytics, 2024)](https://www.sciopen.com/article/10.26599/BDMA.2024.9020031)\n29. [Synchrosqueezing transforms: From low- to high-frequency modulations and perspectives (Comptes Rendus Physique, 2019)](https://comptes-rendus.academie-sciences.fr/physique/item/10.1016/j.crhy.2019.07.001.pdf)\n30. [Adaptive multi-scale TF-net for high-resolution time–frequency representations (Signal Processing)](https://www.sciencedirect.com/science/article/abs/pii/S0165168423003213)\n31. [Spectral estimation, What is new? What is next? (Tary et al., Geophysics)](https://imgw.univie.ac.at/fileadmin/user_upload/i_img/Geophyik/Publikationen_bis_2018/Tary_2014_Spectral.pdf)\n32. [A Quantitative Comparison of Non-Parametric Time-Frequency Representations (EUSIPCO 2005)](https://eurasip.org/Proceedings/Eusipco/Eusipco2005/defevent/papers/cr1246.pdf)\n33. [Time-frequency techniques in biomedical signal analysis: a tutorial review of similarities and differences](https://pubmed.ncbi.nlm.nih.gov/23703538/)\n34. [FBS Perfect Reconstruction (ccrma.stanford.edu)](https://ccrma.stanford.edu/~jos/sasp/FBS_Perfect_Reconstruction.html)\n35. [pmc.ncbi.nlm.nih.gov](https://pmc.ncbi.nlm.nih.gov/articles/PMC10346397/)\n\n---\n*Topic: Encyclopedia › Technology and the built world › Computing and digital systems › Artificial intelligence and data › Algorithms and computational methods › Numerical, string, and geometric algorithms › Fourier and signal transforms*\n\n*Initially written Sep 29, 2026 · Reviewed: Sep 30, 2026 · Edited: Sep 30, 2026 · Last review: Sep 30, 2026*\n\n*Copyright 2026 EdgeChat AI, a subsidiary of Biostate AI.*\n\nLicense: Edgepedia Community License 1.0, https://www.edgepedia.org/edgepedia/license\n", "same_as": [ "https://sites.math.duke.edu/~ingrid/publications/ieee36-1990.pdf", "https://www.math.ucdavis.edu/~saito/data/synchrosqueezing/daubechies-lu-wu_synchrosqueezing.pdf", "https://ccrma.stanford.edu/~jos/sasp/FBS_Perfect_Reconstruction.html" ], "url": "https://www.edgepedia.org/time-frequency-representation", "markdown_url": "https://www.edgepedia.org/time-frequency-representation.md", "license": { "name": "Edgepedia Community License 1.0", "url": "https://www.edgepedia.org/edgepedia/license", "summary": "Free with credit, commercial use included. AI training is open to everyone. For other uses, organizations over USD 100M in revenue or 100M monthly users license separately.", "spdx": "LicenseRef-Edgepedia-Community-1.0" }, "credit": "\"Time–frequency representation\", Edgepedia (EdgeChat), https://www.edgepedia.org/time-frequency-representation. Edgepedia Community License 1.0.", "credit_md": "\"[Time–frequency representation](https://www.edgepedia.org/time-frequency-representation)\", Edgepedia (EdgeChat), [https://www.edgepedia.org/time-frequency-representation](https://www.edgepedia.org/time-frequency-representation). [Edgepedia Community License 1.0](https://www.edgepedia.org/edgepedia/license).", "credit_html": "\"Time–frequency representation\", Edgepedia (EdgeChat), https://www.edgepedia.org/time-frequency-representation. Edgepedia Community License 1.0.", "speakable": "A time–frequency representation is a signal processing method that maps a signal's energy over time and frequency at once, with the spectrogram as the standard example." }