Acoustic Descriptors for Characterization of Musical Timbre Using the Fast Fourier Transform

Gonzalez, Yubiry; Prati, Ronaldo C.

doi:10.3390/electronics11091405

Cited by 6 publications

(9 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Analyzing the frequency response is a common practice when studying the behavior of a musical instrument [ 13 , 23 , 24 , 25 ], comparing different models or the differences between performers.…”

Section: Methodsmentioning

confidence: 99%

Design, Manufacturing and Acoustic Assessment of Polymer Mouthpieces for Trombones

et al. 2023

View full text Add to dashboard Cite

Brass instruments mouthpieces have been historically built using metal materials, usually brass. With the auge of additive manufacturing technologies new possibilities have arisen, both for testing alternative designs and for using new materials. This work assesses the use of polymers for manufacturing trombone mouthpieces, specifically PLA and Nylon. The acoustical behavior of these two mouthpieces has been compared with the obtained from a third one, built from brass. Both additive and subtractive manufacturing techniques were used, and the whole manufacturing process is described. The mouthpieces were acoustically assessed in an anechoic chamber with the collaboration of a professional performer. The harmonic analysis confirmed that all the manufactured mouthpieces respect the harmonic behavior of the instrument. An energy analysis of the harmonics revealed slight differences between the mouthpieces, which implies differences in the timbre of the instrument. Although these subtle differences would not be acceptable when performing with the instrument in an orchestra, they could be perfectly valid for early learners, personal rehearsals or any kind of alternative performance.

show abstract

Section: Methodsmentioning

confidence: 99%

Design, Manufacturing and Acoustic Assessment of Polymer Mouthpieces for Trombones

et al. 2023

View full text Add to dashboard Cite

show abstract

“…Recent studies show that even deep learning models can achieve better recognition accuracy if meaningful features are extracted. Among feature extraction techniques, FFT has proven highly effective [ 70 , 71 , 72 , 73 , 74 , 75 , 76 , 77 , 78 , 79 , 80 ], including on the DEAP [ 48 ]. In this study, FFT was chosen to extract features from DEAP EEG signals.…”

Section: Methodsmentioning

confidence: 99%

M1M2: Deep-Learning-Based Real-Time Emotion Recognition from Neural Activity

Akter

Prodhan

Pias

et al. 2022

Sensors

View full text Add to dashboard Cite

Emotion recognition, or the ability of computers to interpret people’s emotional states, is a very active research area with vast applications to improve people’s lives. However, most image-based emotion recognition techniques are flawed, as humans can intentionally hide their emotions by changing facial expressions. Consequently, brain signals are being used to detect human emotions with improved accuracy, but most proposed systems demonstrate poor performance as EEG signals are difficult to classify using standard machine learning and deep learning techniques. This paper proposes two convolutional neural network (CNN) models (M1: heavily parameterized CNN model and M2: lightly parameterized CNN model) coupled with elegant feature extraction methods for effective recognition. In this study, the most popular EEG benchmark dataset, the DEAP, is utilized with two of its labels, valence, and arousal, for binary classification. We use Fast Fourier Transformation to extract the frequency domain features, convolutional layers for deep features, and complementary features to represent the dataset. The M1 and M2 CNN models achieve nearly perfect accuracy of 99.89% and 99.22%, respectively, which outperform every previous state-of-the-art model. We empirically demonstrate that the M2 model requires only 2 seconds of EEG signal for 99.22% accuracy, and it can achieve over 96% accuracy with only 125 milliseconds of EEG data for valence classification. Moreover, the proposed M2 model achieves 96.8% accuracy on valence using only 10% of the training dataset, demonstrating our proposed system’s effectiveness. Documented implementation codes for every experiment are published for reproducibility.

show abstract

“…For each recording, the FFT is obtained with normalized amplitudes, using the SciPy library module in Python [11]. Timbral coefficients are calculated from the FFTs, which are dimensionless, univocal, and independent descriptors of the FFTs [6,8]. These six timbral coefficients, together with the fundamental frequency (f 0 ) provide, for each audio record, a seven-dimensional vector that defines a point in an abstract space, which also is a geometric space.…”

Section: Methodsmentioning

confidence: 99%

“…For a specific musical instrument, the relative measure of the amplitude of the fundamental frequency with respect to the set of amplitudes of the FFT (Affinity Coefficient A) and the average variation of the envelope of the pulses in the FFT (Monotonicity coefficient M) are associated to the musical octave [8], the difference in the composition of harmonics (Spectral Signature) and the average value of the harmonicity of the partial frequencies (Harmonicity coefficient H) allow to identify the musical instrument [6]. For a given musical instrument and a specific musical sound, the relative measure of the amplitude of the fundamental frequencies (Sharpness coefficient S, note that this is not Zwicker's psychoacoustic sharpness) and the average of the deviation of the amplitudes of the partial frequencies with respect to the amplitude of the fundamental (MA Coefficient) report dynamics [8]. However, different musical sounds played by different instruments can be perceived as timbrically similar, and therefore should be close in timbral space [7].…”

Section: Methodsmentioning

confidence: 99%

Musical timbre classification using FFT-Acoustic Descriptors and Machine Learning

Gonzalez,

Prati

2022

Proceedings of the 10th Convention of the European Acoustics Association Forum Acusticum 2023

View full text Add to dashboard Cite

Musical timbre is a complex multidimensional attribute of auditory perception, which allows, in a first approximation, to discriminate between musical instruments when they have the same sound, intensity, and duration. Also, in some cases, there are sounds that appear to have very close timbral similarity, even when the instruments have different acoustic characteristics. This fact can make it difficult to classify musical instruments by timbres. We explore a 7dimensional abstract space, formed by the fundamental frequency and acoustic descriptors extracted from Fourier Transform in five musical instruments: Trumpet, violin, cello, transverse flute, and clarinet, of a monophonic audio record, from the Tinysol and Good-Sounds databases, corresponding to the fourth octave. This approach makes it possible to define a collection of points in timbral space uniquely and allows differentiating sounds played in ordinary style on any type of musical instrument. Through the geometric distance between musical sounds, we explore some Machine Learning techniques to establish categories of similarities between musical sounds, instruments, and family of musical instruments. It is concluded that the study of timbral similarity through geometric distances made it possible to find clustering between categories of musical timbre.

show abstract

Acoustic Descriptors for Characterization of Musical Timbre Using the Fast Fourier Transform

Cited by 6 publications

References 25 publications

Design, Manufacturing and Acoustic Assessment of Polymer Mouthpieces for Trombones

Design, Manufacturing and Acoustic Assessment of Polymer Mouthpieces for Trombones

M1M2: Deep-Learning-Based Real-Time Emotion Recognition from Neural Activity

Musical timbre classification using FFT-Acoustic Descriptors and Machine Learning

Contact Info

Product

Resources

About