|
|
Electric Guitar to MIDI Conversion
Klčo, Michal ; Glembek, Ondřej (referee) ; Černocký, Jan (advisor)
Automatický přepis hudby a odhad vícero znějících tónu jsou stále výzvou v oblasti dolování informací z hudby. Moderní systémy jsou založeny na různých technikách strojového učení pro dosažení co nejpřesnějšího přepisu hudby. Některé z nich jsou také omezeny na konkrétní hudební nástroj nebo hudební žánr, aby se snížila rozmanitost analyzovaného zvuku. V této práci je navrženo, vyhodnoceno a porovnáváno několik systémů pro konverzi nahrávek elektrické kytary do MIDI souború, založených na různých technikách strojového učení a technikách spektrální analýzy.
|
|
|
Extraction of parameters for the research of music performance
Laborová, Anna ; Miklánek, Štěpán (referee) ; Ištvánek, Matěj (advisor)
Different music performances of the same piece may significantly differ from each other. Not only the composer and the score defines the listener’s music experience, but the music performance itself is an integral part of this experience. Four parameter classes can be used to describe a performance objectively: tempo and timing, loudness (dynamics), timbre, and pitch. Each of the individual parameters or their combination can generate a unique characteristic performance. The extraction of such objective parameters is one of the difficulties in the field of Music Performance Analysis and Music Information Retrieval. The submitted work summarizes knowledge and methods from both of the fields. The system is applied to extract data from 31 string quartet performances of 2. movement Lento of String Quartet no. 12 F major (1893) by czech romantic composer Antonín Dvořák (1841–1904).
|
|
|
Web interface for audio feature visualization
Putz, Viliam ; Ištvánek, Matěj (referee) ; Miklánek, Štěpán (advisor)
This thesis deals with methods of audio features extraction from audio files, visualization of these features and implementation of web interface, which provides the visualization. In the introduction, Music Information Retrieval field, with which this thesis is closely related, is described. Also, the current state in the area of applications for audio features extraction is described. Next, the most common libraries for audio feature extraction within the programming languages are listed. In the second chapter, the audio features that can be extracted from audio file are listed and described. In the third chapter, there is description of the process of implementation, used technologies, function diagram of the web interface, explanation of functionality and description of user interface and its functions.
|
|
|
Recognizing the historical period of interpretation based on the music signal parameterization
Král, Vítězslav ; Mucha, Ján (referee) ; Kiska, Tomáš (advisor)
The aim of this semestral work is to summarize the existing knowledge from the area of comparison of musical recordings and to implement an evaluation system for determining the period of creation using the music signal parameterization. In the first part of this work are describe representations which can music take. Next, there is a cross-section of parameters that can be extracted from music recordings provides information on the dynamics, tempo, color, or time development of the music’s recording. In the second part is described evaluation system and its individual sub-blocks. The input data for this evaluation system is a database of 56 sound recordings of the first movement of Beethoven’s 5th Symphony. The last chapter is dedicated to a summary of the achieved results.
|
|
|
System for finding duplicate recordings based on audio information
Švejcar, Michael ; Miklánek, Štěpán (referee) ; Ištvánek, Matěj (advisor)
This diploma thesis discusses different methods of detecting duplicates in a music file database. The problem at hand is that files containing the same recording may differ in sound quality, applause at the end of a performance and other such parameters. The aim of this thesis is to design and implement a system that identifies duplicate recordings and provides an output file for the comparison. The system needs to not be affected by the mentioned parameters but precise enough to prevent matching non-identical recordings. The system is realized using the Python programming language, freely available libraries for computing chroma features, Image Hashing technique and multiple variants of the dynamic time warping algorithm. Three comparison methods were implemented in the system, differing in precision and computation complexity. The methods were then tested on a prepared dataset and four preset precision options were created. The final system seems very precise and insusceptible to detecting recordings that are very similar but not identical as duplicates, for example in case of different interpretations of the same musical piece.
|
|
|
Automatic tagging of musical compositions using machine learning methods
Semela, René ; Galáž, Zoltán (referee) ; Kiska, Tomáš (advisor)
One of the many challenges of machine learning are systems for automatic tagging of music, the complexity of this issue in particular. These systems can be practically used in the content analysis of music or the sorting of music libraries. This thesis deals with the design, training, testing, and evaluation of artificial neural network architectures for automatic tagging of music. In the beginning, attention is paid to the setting of the theoretical foundation of this field. In the practical part of this thesis, 8 architectures of neural networks are designed (4 fully convolutional and 4 convolutional recurrent). These architectures are then trained using the MagnaTagATune Dataset and mel spectrogram. After training, these architectures are tested and evaluated. The best results are achieved by the four-layer convolutional recurrent neural network (CRNN4) with the ROC-AUC = 0.9046 ± 0.0016. As the next step of the practical part of this thesis, a completely new Last.fm Dataset 2020 is created. This dataset uses Last.fm and Spotify API for data acquisition and contains 100 tags and 122877 tracks. The most successful architectures are then trained, tested, and evaluated on this new dataset. The best results on this dataset are achieved by the six-layer fully convolutional neural network (FCNN6) with the ROC-AUC = 0.8590 ± 0.0011. Finally, a simple application is introduced as a concluding point of this thesis. This application is designed for testing individual neural network architectures on a user-inserted audio file. Overall results of this thesis are similar to other papers on the same topic, but this thesis brings several new findings and innovations. In terms of innovations, a significant reduction in the complexity of individual neural network architectures is achieved while maintaining similar results.
|
|
|
Beat Tracking: Is 441 kHz Really Needed?
Ištvánek, Matěj ; Miklánek, Štěpán
Beat tracking is essential in music informationretrieval, with applications ranging from music analysis and automaticplaylist generation to beat-synchronized effects. In recentyears, deep learning methods, usually inspired by well-knownarchitectures, outperformed other beat tracking algorithms. Thecurrent state-of-the-art offline beat tracking systems utilize temporalconvolutional and recurrent networks. Most systems use aninput sampling rate of 44.1 kHz. In this paper, we retrain multipleversions of state-of-the-art temporal convolutional networks withdifferent input sampling rates while keeping the time resolutionby changing the frame size parameter. Furthermore, we evaluateall models using standard metrics. As the main contribution,we show that decreasing the input audio recording samplingfrequency up to 5 kHz preserves most of the accuracy, and insome cases, even slightly outperforms the standard approach.
|
|
|
Piano chord analyzer
Poloček, Dominik ; Miklánek, Štěpán (referee) ; Ištvánek, Matěj (advisor)
The presented thesis deals with the analysis of chords by determining the frequencies of their components. The aim of thesis is to outline methods for determining the fundamental frequencies of single and multiple notes and to implement a system that can determine chords using these methods. The method, implemented in Python (spectral peak method), uses a fast Fourier transform to represent the signal in the frequency domain and then searches for spectral maxima, which it evaluates as fundamental frequencies after proper checking. The spectral peaks method was compared with the harmonic component modulus summation method and with the state-of-the- art system for transcribing recordings to MIDI (PianoTransctiprion) by running tests on the dataset created for this thesis (530 chord and note recordings). The best results are presented by PianoTranscription ( = 0.74, tot = 0.23), the second best performing method is the spectral peaks method with a known number of tones ( = 0.55, tot = 0.29), followed by the same method with unknown number of tones ( = 0.52, tot = 0.38) and finally the harmonic component modulus summation method ( = 0.26, tot = 0.81). The limitations of the implemented system are the inability to determine the number of tones (must be specified by the user) and the frequency minimum (138.59 Hz), below which the estimates are erroneous, which is probably due to the design of the piano and the braiding of strings.
|
|
|
Analysis of automatic parameter extraction on piano recordings
Kaplan, Josef ; Miklánek, Štěpán (referee) ; Ištvánek, Matěj (advisor)
This bachelor thesis deals with the analysis of the accuracy of automatic extraction of parameters, mainly of piano recordings. The given issue is described both from a technical and a musical perspective. This thesis summarizes knowledge from the field of music theory and the automatic detection of parameters that can be obtained from musical piano recordings. This thesis is focused on detecting onsets, beats, downbeats, pitch estimation and tempo. The analysis of piano recordings is realized using the Python programming language. The output is scripts that perform parameter detection based on user-selected methods that are commonly used to calculate parameters. The result is also testing the accuracy of individual methods based on annotations from different datasets, focusing primarily on piano recordings. The final part contains an evaluation based on selected metrics with an objective comparison.
|
|
|
Web application for visualization of music recording parameters
Klimeš, Martin ; Ištvánek, Matěj (referee) ; Miklánek, Štěpán (advisor)
This thesis focuses on the development of a web application for visualizing musical parameters. The goal is to provide users with an environment where they can easily visualize parameters of any music recording and compare these parameters across different interpretations of the composition. The musical parameters visualized in the application are based on the field of Music Information Retrieval. For each of these visualizations, the application implements various settings that are saved to a database for the loggedin user, allowing them to adjust the visualization display according to their individual needs. The reactive Vue.js framework was used for the client-side, Flask framework for the server-side, and the PostgreSQL relational database system for data storage.
|
guest ::
