|
|
Localization and Rendering of Sound Sources in Acoustic Fields
Khaddour, Hasan ; Rakús, Martin (oponent) ; Husník, Libor (oponent) ; Schimmel, Jiří (vedoucí práce)
This doctoral thesis deals with sound source localization and acoustic zooming. The primary goal of this dissertation is to design an acoustic zooming system, which can zoom the sound of one speaker among multiple speakers even when they speak simultaneously. The system is compatible with surround sound techniques. In particular, the main contributions of the doctoral thesis are as follows: 1. Design of a method for multiple sound directions estimations. 2. Proposing a method for acoustic zooming using DirAC. 3. Design a combined system using the previous mentioned steps, which can be used in teleconferencing.
|
|
|
Modern Methods of Time-Frequency Warping of Sound Signals
Trzos, Michal ; Rakús, Martin (oponent) ; Průša, Zdeněk (oponent) ; Schimmel, Jiří (vedoucí práce)
This thesis deals with representation of non-stationary harmonic signals with time-varying components. Its main focus is aimed at Harmonic Transform and its variant with subquadratic computational complexity, the Fast Harmonic Transform. Two algorithms using the Fast Harmonic Transform are presented. The first uses the gathered log-spectrum as fundamental frequency change estimation method, the second uses analysis-by-synthesis approach. Both algorithms are used on a speech segment to compare its output. Further the analysis-by-synthesis algorithm is applied on several real sound signals to measure the increase in the ability to represent real frequency-modulated signals using the Harmonic Transform.
|
|
|
Tool for Effective Communication between a Drone Pilot and a Mission Commander
Rakús, Martin ; Zemčík, Pavel (oponent) ; Bambušek, Daniel (vedoucí práce)
Nowadays, the primary communication channel for firefighters and mountain rescue units is voice transferred over the radio. However, radio communication might not be sufficient enough with the increasing use of drones in rescue operations, especially for communication between drone pilots and mission commanders or operators. When relying solely on radio communication, the ground team and the commander have to depend on verbal communication alone, which can lead to restricted communication and the possibility of missed or misinterpreted instructions. To address this, I developed two applications - one for the drone pilot and one for the unit commander. The pilot app includes the interface necessary for flying the drone, while the commander app connects to the pilot app and provides access to the drone video feed and other important information. The commander can send visual feedback to the pilot by drawing into the video feed, placing annotation icons, or sending flight commands for more precise cooperation between the drone and the ground team. As the solution is a stand-alone system, it is not susceptible to radio communication overload, which guarantees that commands are not lost and that it can be used even when the radio fails.
|
|
|
Localization and Rendering of Sound Sources in Acoustic Fields
Khaddour, Hasan ; Rakús, Martin (oponent) ; Husník, Libor (oponent) ; Schimmel, Jiří (vedoucí práce)
This doctoral thesis deals with sound source localization and acoustic zooming. The primary goal of this dissertation is to design an acoustic zooming system, which can zoom the sound of one speaker among multiple speakers even when they speak simultaneously. The system is compatible with surround sound techniques. In particular, the main contributions of the doctoral thesis are as follows: 1. Design of a method for multiple sound directions estimations. 2. Proposing a method for acoustic zooming using DirAC. 3. Design a combined system using the previous mentioned steps, which can be used in teleconferencing.
|
|
|
Modern Methods of Time-Frequency Warping of Sound Signals
Trzos, Michal ; Rakús, Martin (oponent) ; Průša, Zdeněk (oponent) ; Schimmel, Jiří (vedoucí práce)
This thesis deals with representation of non-stationary harmonic signals with time-varying components. Its main focus is aimed at Harmonic Transform and its variant with subquadratic computational complexity, the Fast Harmonic Transform. Two algorithms using the Fast Harmonic Transform are presented. The first uses the gathered log-spectrum as fundamental frequency change estimation method, the second uses analysis-by-synthesis approach. Both algorithms are used on a speech segment to compare its output. Further the analysis-by-synthesis algorithm is applied on several real sound signals to measure the increase in the ability to represent real frequency-modulated signals using the Harmonic Transform.
|
host ::
RSS.