Read the abstract, see who wrote it and who cites it, and get the citation in APA, MLA, Chicago or Harvard.
About 10,000 papers for “Spectrogram”
Katie Campbell (2498896), Yelena Cerezke-Riemer (21169987), John H. Acorn (4859542) · Figshare · 2025
A macrophyte bubble stream sound recorded August 1, 2021, at Gull Lake, Alberta, Canada. See <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0317424#pone.0317424.g003" target="_blank">Fig 1</a> for axis descriptions.
Cited by 0Open access
W. Ross Silcock (11372246), Shari L. Schwartz (11372249), John U. Carlini (11372252) et al. · Figshare · 2021
<p>Two examples of each of the three subcategories of songs not readily assignable as Type A, B, or C have been selected from <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0256385#pone.0256385.s001" target="_blank">S1 Appendix</a>. The analyzed introductory (1) and unanalyzed secondary (2) note sequences are designated by brackets; classifications are determined solely by the introductory note sequence when practicable.</p>
Cited by 0Open access
Jan Krajíček · Digital Repository (National Repository of Grey Literature) · 2010
The thesis describes generation of time-frequency-intensity graphs (spectrograms) of audio recordings using the Fourier transform. The theoretical possibilities and limitations of spectrogram synthesis (audio reconstruction of a given spectrogram) are discussed and two practical synthesis methods are described, based on reconstruction using pure tones and random noise. Spectrogram generation and both described synthesis methods are implemented in the form of a computer program with a graphical user interface, allowing straightforward configuration of relevant parameters.
Cited by 0
Coluccelli, Nicola · Zenodo (CERN European Organization for Nuclear Research) · 2023
A) Standoff CARS spectrogram of B. atrophaeus spores. Pump, Stokes, and probe pulse energies are 12, 12, and 9 μJ, respectively. The spore sample is placed at 1 m from the laser system and the integration time is 1 s. (B,D) Cross-sections (solid lines) of standoff CARS spectrogram of B. atrophaeus spores and ALS fitting profiles (dashed lines) for probe delays of 0.5 and 1.5 ps. (C,E) Corresponding residuals and reference Raman spectrum of CaDPA (magenta line). Residuals represent broadband CARS fingerprints of B. atrophaeus spores.
Cited by 0Open access
Krzysztof Czarnecki · Mechanical Systems and Signal Processing · 2015
Cited by 38
Josef Fruehwald · Front Matter · 2023
I might flesh this out as a more detailed tutorial for LingMethodsHub, but for now this is going to be a rough-around-the-edges post about making spectrograms in R. My goal will be to get as close as possible to recreating a spectrogram like you might get from Praat. Pre-processing.
Cited by 0Open access
Sreyan Ghosh, Ashish Seth, S. Umesh et al. · 2023
We present Multiscale Audio Spectrogram Transformer (MAST) for audio classification, which brings the concept of multiscale feature hierarchies to the Audio Spectrogram Transformer (AST) [1]. Given an input audio spectrogram, we first patchify and project it into an initial temporal resolution and embedding dimension, post which the multiple stages in MAST progressively expand the embedding dimension while reducing the temporal resolution of the input. We use a pyramid structure that allows early layers of MAST operating at a high temporal resolution but low embedding space to model simple low-level acoustic information and deeper temporally coarse layers to model high-level acoustic information with high-dimensional embeddings. We also extend our approach to present a new Self-Supervised Learning (SSL) method called SS-MAST, which calculates a symmetric contrastive loss between latent representations from a student and a teacher encoder, leveraging patch-drop, a novel audio augmentation approach that we introduce. In practice, MAST significantly outperforms AST by an average accuracy of 3.4% across 8 speech and non-speech tasks from the LAPE Benchmark [2], achieving state-of-the-art results on keyword spotting in Speech Commands. Additionally, our proposed SS-MAST achieves an absolute average improvement of 2.6% over the previously proposed SSAST [3]<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sup>.
Cited by 4
Josef Fruehwald · Front Matter · 2023
I might flesh this out as a more detailed tutorial for LingMethodsHub, but for now this is going to be a rough-around-the-edges post about making spectrograms in R. My goal will be to get as close as possible to recreating a spectrogram like you might get from Praat. Pre-processing.
Cited by 0Open access
Anugunj Naman, G.J. Zhang · 2025
In audio classification, developing efficient and robust models is critical for real-time applications. Inspired by the design principles of MobileViT, we present FAST (Fast Audio Spectrogram Transformer), a new architecture that combines convolutional neural networks (CNNs) and transformers to capitalize on the strengths of both. FAST integrates the local feature extraction efficiencies of CNNs with the global context modeling capabilities of transformers, resulting in a model that is powerful yet lightweight, well-suited to a real-time or mobile use case. Additionally, we incorporate Lipschitz continuous attention mechanisms to improve training stability and accelerate convergence. We evaluate FAST on the ADIMA dataset, a multilingual corpus towards real-time profanity and abuse detection, as well as on the more traditional AudioSet. Our results show that FAST achieves state-of-the-art performance on both the ADIMA and AudioSet classification tasks and in some cases surpasses existing benchmarks while using up to 150x fewer parameters.
Cited by 3
T.N.S. Tengku Zawawi, Abdul Rahim Abdullah, Ezreen Farina Shair et al. · 2013
Electromyography (EMG) is known as complex bioelectricity signals that representing the contraction of the muscle in humanbody. The EMG signal offers useful information that can help to understand the human movement. Many techniques have been proposed by various researchers such as fast Fourier transforms (FFT). However, the technique only gives temporal information of the signal and does not suitable for EMG that consists of magnitude and frequency variation. In this paper, the analysis of EMG signal is presented using time-frequency distribution (TFD) which is spectrogram with different window size. Since the spectrogram represent the the EMG signal in time-frequency representation (TFR), it is very appropriate to analyze the signal. The EMG signals from Biceps muscle of two subjects are collected for body position of 0° and 90°. From the TFR, parameters of the signal such as instantaneous fundamental root mean square (RMS) voltage (Vrms) are estimated. To identify the suitable windows size, spectrogram with size window of 64, 256, 512 and 1024 is used to analyze the signal and the performance of the TFR are evaluated. The results show that spectrogram with window size of 512 gives optimal TFR of the EMG signals and suitable to characterize the signal.
Cited by 24
Monstein, Christian, Csillaghy, André, Benz, Arnold O. · Space Physics Data Facility · 2023
This dataset contains solar dynamic spectrogram FITS files of the CALLISTO spectrometer data from the e-Callisto network of stations. From the website http://e-callisto.org - The CALLISTO spectrometer is a programmable heterodyne receiver built in the framework of IHY2007 and ISWI by former Radio and Plasma Physics Group (PI Christian Monstein) at ETH Zurich, Switzerland. The main applications are observation of solar radio bursts and rfi-monitoring for astronomical science, education and outreach. Many CALLISTO instruments have been deployed. A complete list of stations is available from the website.
Cited by 5Open access
Ana Guerrero-Tamayo (19484930), Borja Sanz Urquijo (19484933), Isabel Olivares (478244) et al. · Figshare · 2024
<p>In the case of HIV-1, the length of the x-axis is approximately 10,000 nucleotides.</p>
Cited by 0Open access
Laura A. Brooks, M. Kidner, Anthony C. Zander et al. · Adelaide Research & Scholarship (AR&S) (University of Adelaide) · 2006
The determination of seabed properties and their incorporation in sound propagation models for estimating the effectiveness of sonar and underwater communications is generally required in shallow water regions in the vicinity of commonly used shipping and fishing routes. It has been suggested by others that the extraction of acoustic data from interference patterns observable in ambient noise spectrograms can be used as an effective means of determining seafloor properties in these situations. Within this paper, the problem of obtaining information from a frequency vs. range spectrogram and relating this to the governing modal model is discussed. An existing method for estimating modal group and phase velocities (and hence seabed properties) from spectrogram information, based upon a dual mode analysis, is critically reviewed and the need for a more complete description of the modal model is outlined.
Cited by 8
Hongqiang Chai (13170571), Guolai Yang (13170574), Guoguo Wu (13170577) et al. · Figshare · 2022
Cited by 0Open access
A. F. M. Shahen Shah · 2025
In this study, we explore the use of Vector Quantized Variational Autoencoders (VQ-VAE) for real-time audio spectrogram inpainting, with a focus on minimizing environmental impact. We leverage the capabilities of VQ-VAE, a generative model that compresses and quantizes spectrograms into compact code maps, to regenerate missing or corrupted parts of audio signals. Our experiments, conducted on both the MNIST image dataset and the NSynth audio dataset, demonstrate the model’s effectiveness in reconstructing spectrograms. We also introduce a real-time CO2 tracker to monitor the energy consumption and carbon footprint of our training processes. Our results highlight the potential of VQ-VAE in the field of computer music, providing innovative tools for musicians while addressing the environmental costs of machine learning.
Cited by 1
Frances Ingemann, P. Mermelstein · The Journal of the Acoustical Society of America · 1975
In order to assess human analysis of acoustic data before attempting such analysis by machine, a series of experiments was conducted in which subjects were asked to match spectrograms of continuous speech to reference spectrograms of the same words. Although error rates varied with sentence difficulty and size of vocabulary, comparison of the matches shows greater agreement in phoneme segments than other experimenters have obtained in phonetic transcriptions of unknown utterances without semantic of syntactic processing. Accuracy in matching can be further improved by feedback to the human analyst in the form of spectrographic representation of a sequence of tentative matches spoken as if they made up the unknown utterance. The results suggest that automatic matching of word− or syllable−sized acoustic patterns may provide a more accurate phonemic input to the syntactic−semantic component of a speech recognition system than direct transcription of individual phonetic segments. Subject Classification: 70.65, 70.60.
Cited by 3
Alessandro De Luca, Micaela Contu, Stanislav Hristov et al. · 2016
This paper focuses on the estimation of target motion parameters in forward scatter radar. A new method based on the use of the Doppler signature spectrogram has been investigated. This allows a rough estimation of target motion parameters even in the presence of significant clutter, when the Doppler signature cannot be used to extract accurate kinematic information. The spectrogram characteristics change depending on target's speed, crossing angle and crossing point, this will be shown in order to explain how it is possible to estimate kinematic information analyzing the spectrogram trend. The performance of this new method is analyzed at the initial stage of the research.
Cited by 24
Donald A. Gurnett · Open MIND · 2026
This dataset contains dynamic spectrogram PNG plots of the ISEE1/PWI Sweep Frequency Receiver (SFR) data. Each plot spans one day. These data were obtained from the Ev long wire antenna in the spin plane and the Eu two-sphere antenna via the SFR receiver. The title above each spectrogram is "ISEE-1 PWI Sweep Frequency Receiver Ev and Eu Antennas" Each spectrogram is a plot of the power spectral density of received signal (color scale) as a function of operating frequency (in a logarithmic scale on the vertical axis) and time (horizontal axis). Beneath the time labels on the horizontal axis of the spectrograms are ephemeris data: position of the spacecraft in radial distance (Earth radii), geomagnetic latitude, magnetic local time, and McIlwain L-shell. Overlaid on each image are traces of the electron cyclotron frequencies. Located in the same directory are are the 12 hour ISEE-1 PWI Spectrum Analyzer (SA) spectrograms.
Cited by 0Open access
OBrien, Jake Patrick · Digital Commons @ Connecticut College (Connecticut College) · 2021
This internship was about sound post-production in the television industry. I learned how to use Pro Tools in order to create assemblies, edit dialogue, cut sound effects and more. I made several career connections and gained a lot of insight into the post-production industry.
Cited by 0
Quoc Bao Diep (18452207), Hong Yen Phan (18452210), Thanh-Cong Truong (18452213) · Figshare · 2024
<div><p>Digital speech recognition is a challenging problem that requires the ability to learn complex signal characteristics such as frequency, pitch, intensity, timbre, and melody, which traditional methods often face issues in recognizing. This article introduces three solutions based on convolutional neural networks (CNN) to solve the problem: 1D-CNN is designed to learn directly from digital data; 2DS-CNN and 2DM-CNN have a more complex architecture, transferring raw waveform into transformed images using Fourier transform to learn essential features. Experimental results on four large data sets, containing 30,000 samples for each, show that the three proposed models achieve superior performance compared to well-known models such as GoogLeNet and AlexNet, with the best accuracy of 95.87%, 99.65%, and 99.76%, respectively. With 5-10% higher performance than other models, the proposed solution has demonstrated the ability to effectively learn features, improve recognition accuracy and speed, and open up the potential for broad applications in virtual assistants, medical recording, and voice commands.</p></div>
Cited by 0Open access