M.Sc. Carlos A. de Obaldía P.

Carlos de Obaldia

Room:
118
Telephone:
+49-40-6541-2870
Fax:
+49-40-6541-2834
Email:
[email protected]
Address (visitors)
Helmut-Schmidt-Universität
Building H3
Holstenhofweg 85
22043 Hamburg
Address
Helmut-Schmidt-Universität
Fakultät für Elektrotechnik
Allgemeine Nachrichtentechnik
Postfach 70 08 22
22008 Hamburg

 

Research

Passive Acoustic Monitoring

Abundance Estimation for Passive Acoustic Monitoring (PAM) of Cetaceans
Several algorithms for acoustic monitoring in underwater environments are evaluated for real-time classification of call signals from marine mammals, in particular of cetaceans.
System for Click Detection

Voice Conversion

Voice conversion based on a phonetic reconstruction of speech
Speech can be discretized in voiced and unvoiced regions. Voiced regions of speech mainly contain a specific phoneme or transitions between particular phonemes. Parts of speech can then be abstracted to their language-independent phonetic representation and saved for further processing. Real-time phonetic tagging is applied to an incoming speech waveform from a source speaker. These phonetic segments are exchanged with those of a particular target speaker and synthesized in the back-end, thus converting speech at the phonetic level.
Speech Morphing

3D Model Reconstruction

3D model reconstruction based on 2D images
Object contours and camera positions are traced in 2-D images. These features are then used for a volumetric reconstruction of objects, particularly ships.  Several performance indicators are used to asses the quality of the results. Innovative and known methods in volumetric reconstruction and image segmentation are evaluated to design an optimized framework for 3D model reconstruction.

Publications

C. de Obaldía, U. Zölzer: “Vowel Conversion by Phonetic Segmentation”, in Proc. of the 18th International Conference on Digital Audio Effects (DAFx-15),  Nov. 30-Dec. 3 2015, Trondheim, Norway

C. de Obaldía, G. Simkus, U. Zölzer: “Estimating the number of sperm whale (Physeter macrocephalus) individuals based on grouping of corresponding clicks”, in Proc. of the 41th German Annual Conference on Acoustics (DAGA 2015),  March. 16-19 2015, Nurenberg, Germany

C. de Obaldía,  U. Zölzer: “Estimation of the number of whale individuals based on click sounds of selected whale species”, in Proc. of the 43rd German Annual Conference on Acoustics (DAGA 2017),  March. 06-09 2017, Kiel, Germany

Supervised Thesis

Yicheng Peng: “Ship Detection, Extraction and Classification based on Satellite Images”, Studienarbeit, August 2014

Chen Cheng: “Space Carving for 3D Ship Model Building”, Studienarbeit, March 2015

Mahesh Sulugodu Manjunatha: “Orthogonal Modeling for 3D Reconstruction of Ships”, Studienarbeit, June 2015

Dennis Muller: “Voiced/Unvoiced Segmentation for Subphonetic Modeling of Speech”, Master thesis, November 2015

Aswin Sampath Kumar: “Spectrogram Segmentation for Sweep Whale Calls”, Studienarbeit, September 2016

Sumanth Kabbinale: “Convolutional Neural Networks for Spectrogram Segmentation of Whale Calls”, Master thesis, August 2017

Ankit Ghosh: “Source Separation and Echo Detection for Underwater Whale Sounds”, Master thesis, in progress

Santosh Basavaraju: “3D Modeling from single view ship Silhouttes using CNNs”, Master thesis, in progress

Praveen Krishna Murthy: “Neural Network based Matched Filters for Click Detection”, Sudienarbeit, in progress

HSU

Letzte Änderung: 22. October 2019