news

[June 10, 2014] Data and tools for distant speech recognition
Our paper on building a dataset for distant speech recognition and the corresponding baseline will be presented at Interspeech 2014.

[June 10, 2014] ATHENA database: real voice data in an office environment
Our paper describing a new database of real multichannel recordings in an office environment will be presented at Interspeech 2014. The database will be publicly released in September.

[May 28th, 2014] Multichannel voice activity detection and robust speaker localization
We will be presenting our current approach for voice activity detection and speaker localization at Eusipco 2014.

[May 21st, 2014] Multimodal gesture recognition
Our multimodal gesture recognition system outperforms all other participants in the CHALEARN challenge. The system will be presented at ICIP 2014.

[March 12th, 2014] ATHENA's voice activity detection and speaker localization
Our paper on voice activity detection and speaker localization has been accepted for presentation at HSCMA 2014.

[February 3rd, 2014] Multichannel processing for far-field recognition
Our paper on distant keyword spotting and speech recognition for home automation has been accepted for presentation at ICASSP 2014.

[September 5th, 2013] SailAlign-v1.3.0 released
A new version of SailAlign has been relased. It now supports Spanish and can run on OSX. Further, the source code is now shared via github. Check the SailAlign webpage.

[March 4th, 2013] @CVSP
Joined the Computer Vision, Speech Communication and Signal Processing (CVSP) group at the National Technical University of Athens. From now on, I will only be updating my new homepage.

[February 24th, 2012] A citizen again
Completed my military service at the Center for Software Support of the greek army.

[May 1st, 2012] Moving to Greece
Moved to Greece from Los Angeles to perform my compulsory military service starting from May 16th, 2012.

[June 8th, 2012] Towards a Better Understanding of the Human Annotation Process
Our paper studying whether human annotation is mainly based on saliency or on causal integration has been accepted for publication in the Proceedings of Interspeech 2012.

[January 6th, 2012] Emotion Recognition and Behavioral Signal Processing papers accepted for publication in ICASSP 2012
The publications webpage has been updated with information about our recently accepted papers for publication in the Proceedings of ICASSP 2012.

[December 8th, 2011] Couples Behavior Audio-Based Classification and Context-Sensitive Learning for Audiovisual Emotion Classification
Our papers on Audio Based Classification of Couples' Behavior and on Context-Sensitive Emotion Learning got accepted for publication in Speech Communication and in IEEE Transactions on Affective Computing respectively.

[November 12th, 2011] Publications' webpage updated
The publications webpage has been updated to include links to pdf versions of the papers and presentations or posters.

[November 4th, 2011] ASA meeting in San Diego
Presented our joint work with Michael Proctor on the prosodic characterization of audiobook reading styles. Check the presentations page for additional details and to download a copy of the presented poster.

[October 12th, 2011] ACII 2011
Presented our work on behavioral signal processing at the Int'l Conference on Affective Computing and Intelligent Interaction in Memphis. Uploaded the presentation on the use of Multiple Instance Learning for the classification of behavioral observations and the poster on the estimation of vocal entrainment using pca-based similarity metrics.

[August 31st, 2011] Articulatory recognition using real-time MRI data
The webpage on articulatory recognition using real-time MRI data (art_recognition.html) is under construction and has been updated to provide links to data used for the experiments presented at Interspeech 2011.

[July 16th, 2011] Multiple Instance Learning for Classification of Behavioral Observations
The camera-ready version of the paper on Multiple Instance Learning for Behavioral Signal Processing to be presented in ACII-2011 has been uploaded. You may download it from the publications page.

[June 29th, 2011] SailAlign v1.1 release
SailAlign v1.1 has been released. Thanks to many people from Sail and first users that have pointed out a few problematic cases. Forced phonetic and word level alignment (that may optionally run on top of the iterative recognition-based alignment results) work much better around unaligned regions. Out-of-vocabulary words are treated more robustly.

[June 16th, 2011] ACII 2011 notifications
Our work on multiple instance learning for behavioral signal processing will be presented in the International Conference for Affective Computing and Intelligent Interaction (ACII 2011) in Memphis, Tennessee. Check the publications page for the accepted papers.

[May 27th, 2011] Interspeech 2011 notifications
Recent progress of our research in speech production modeling, and behavioral signal processing will be presented in Interspeech 2011 in Florence. Check the publications page for the accepted papers.

[May 23rd, 2011] ASA Meeting in Seattle
Presented "Multipulse LPC articulatory modeling in the Wisconsin X-ray microbeam speech production database" at the 160th Meeting of the Acoustical Society of America in Seattle. The presented poster is available online.

[May 10th, 2011] Machine Learning and Affective Computing
Became a member of the technical committee of the Machine Learning for Affective Computing workshop. More…

[Jan 30th, 2011] Very Large Scale Research in Phonetics
Presented our work on robust long speech-text alignment at the workshop for Very Large Scale Research in Phonetics, VLSRP-2011. I also gave a tutorial on SailAlign.

[Jan 17th, 2011] ICASSP 2011 notifications
Our most recent work on tracking emotional changes using body language and acoustic cues (paper) and on predicting approach and avoidance ratings of interacting dyads (paper) will be presented in ICASSP 2011 in Prague.

[Sep 30th, 2010] Interspeech 2010
Our paper "Automatic classification of married couples' behavior using audio features" [pdf] was selected as best paper in its area and we got an invitation to submit an extended version of the paper to the Speech Communication journal.