A (very brief) presentation of the Speech Signal Processing - - PowerPoint PPT Presentation

A (very brief) presentation of the Speech Signal Processing Laboratory (SSPL)

George P. Kafentzis Post-Doctoral Researcher & Adjunct Lecturer Department of Computer Science University of Crete

UNIVERSITY OF

C R E T E

 External Members:

• Dr Vassilis Tsiaras

Teaching Staff (ΕΔΙΠ) @ EECS, TUC Machine Learning

• Dr Devora Kiagiadaki

MD-ENT, PhD Speech Pathologies

• Dr Yannis Pantazis,

Researcher @ IACM, FORTH Mathematics of Signal Processing & Deep Learning

Speech Signal Processing Laboratory

 Post-doctoral Researchers:

• Dr Nagaraj Adiga – Speech Enhancement

 PhD Students:

• Muhammed Shifas PV – Wavenet-based Speech Enhancement
• Dipjyoti Paul – GAN-based Voice Conversion

UNIVERSITY OF

C R E T E

• Prof. Yannis Stylianou

Head of SSPL Professor & Senior Research Scientist @ Apple UK IEEE Fellow, ISCA Fellow

Signal Processing Dr George Kafentzis

Post-doctoral Researcher Adjunct Lecturer @ CSD

Signal Processing Dr Anna Sfakianaki

Teaching Staff (ΕΔΙΠ) @ CSD

Phonetics  MSc Students:

• Irene Sisamaki – Text to Speech Synthesis in Greek
• Leonidas Bakayannis – GAN-based Speech Enhancement

 BSc Students:

• Anastassis Livanidis – Speech Dereverberation/Enhancement
• Manolis Kelaidis – Perceptual Coding & Advanced Sinusoidal Models
• Ioanna Kanaria – EEG & Speech Coupling
• Alexandra Kalozoumi – Emotion Detection from Speech Signals

 Members:

Research Interests

 Speech Signal Processing  Audio Signal Processing  Machine/Deep Learning for Speech Processing Specifically:

• Statistical Speech Synthesis

UNIVERSITY OF

C R E T E

Speech Signal Processing Laboratory

• Wavenet
• LDMs
• SSDRC
• wSSDRC
• Adaptive

Sinusoidal Models

• Tremor Estimation
• Jitter/Shimmer Estimation

Speech Signal Processing Laboratory

UNIVERSITY OF

C R E T E

Laboratory ( + 2 external servers equipped with SOTA GPUs )

Speech Signal Processing Laboratory

UNIVERSITY OF

C R E T E

Professional recording booth (worth ~20K €)

Speech Signal Processing Laboratory

UNIVERSITY OF

C R E T E

Professional Laboratory for Speech-related Medical Examinations

Some Projects & Collaborations (2009-…)

 Collaboration Agreements with France Telecom (now Orange) [2009-2013]  Collaboration Agreements with Toshiba Research Europe Limited [2012-2017]  ENRICH – EU Project 675324 Marie Curie European Training Network [2016-…]  Collaboration Agreements with Apple Inc. [2018-…]  Strong collaborations with the Institute

• f Computer Science and Institute of Applied

and Computational Mathematics, FO.R.T.H [2018-…] UNIVERSITY OF

C R E T E

Speech Signal Processing Laboratory

 Several Projects from GME & GSRT  Latsis Foundation Projects

Friends around the world (2009-…)

UNIVERSITY OF

C R E T E

Speech Signal Processing Laboratory

Education

SSPL supports the Computer Science Department by offering the following courses:  CS112 – Physics for Engineers

• Mechanics, Oscillations and Waves, Electromagnetism

 CS215 – Signals and Systems

• Continuous-time Signals, Systems, and Transforms

 CS370 – Digital Signal Processing

• Discrete-time Signals, Systems, and Transforms

 CS590.74 – Introduction to Speech Science and Technologies

• Speech production and perception, phonetics, phonology, etc.

 CS578 – Digital Speech Signal Processing

• Speech production, modeling, analysis,

synthesis, coding, speaker identification, etc.

UNIVERSITY OF

C R E T E

Speech Signal Processing Laboratory

Graduate Courses Undergraduate Courses

Education

SSPL (with the support of the department) organizes a summer school on speech processing each year

Speech Processing Courses in Crete (SPCC) http://www.csd.uoc.gr/~spcc

The Speech Processing Courses in Crete (SPCC) are targeting to teach graduate students and researchers the latest advance- ments of speech processing covering theory, hands-on sessions, and establishing contacts between the academics and industry.

UNIVERSITY OF

C R E T E

Speech Signal Processing Laboratory

Alumni

• Yannis Agiomyrgiannakis, Researcher @ Google, UK (until recently)
• Andre Holzapfel, Assistant Professor @ KTH, Sweden
• Maria Koutsogiannaki, AI Researcher @ Sherpa AI, Spain
• Yannis Pantazis, Researcher @ IACM-FORTH, Greece
• Maria Markaki, Post-doctoral Researcher @ UoC, Greece
• Olina Simantiraki, Ph. D student @ University of Basque Country, Spain
• Veronica Morfi, Ph. D student @ Queen Mary Univ. College, UK
• Miltiadis Vasilakis, Partner & Software Engineer @ Koomasi, Greece
• Maria Astrinaki, Senior Software Engineer @ Sound United, Switzerland
• Myron Apostolakis, Software Engineer @ Sunlight.io, UK
• Theodora Giakoumaki, Software Engineer @ Tom Sawyer Software, Greece
• and more…

UNIVERSITY OF

C R E T E

Speech Signal Processing Laboratory

Selected Publications (2015-…)

• M. Shifas PV, C. Chermaz, T. Chimona, V. Tsiaras, and Y. Stylianou, “Benefits of the WaveNet-Based Speech Intelligibility Enhancement for

Normal and Hearing Impaired Listeners”, In ICA proceeding, 2019.

• M. Shifas PV, C. Santelli, and Y. Stylianou, “Towards Neural-Based Single Channel Speech Enhancement for Hearing Aids”, ICA 2019.
• A. Sfakianaki, “Designing a Modern Greek sentence corpus for audiological and speech technology research”, ICGL14, 2019.
• K. Nicolaidis, A. Sfakianaki, G. Vlahavas, G. P. Kafentzis, “An Acoustic Study of Greek Voiceless Stops”, International Congress of Phonetic

Sciences, Australia, 2019.

• D. Paul, Y. Pantazis, Y. Stylianou, “Non-Parallel Voice Conversion Using Weighted Generative Adversarial Networks”, INTERSPEECH, 2019.
• D. Paul, Y. Pantazis, Y. Stylianou, “Weighted Generative Adversarial Network for many-to-many Voice Conversion”, ICA 2019.
• N. Adiga, Y. Pantazis, V. Tsiaras, and Y. Stylianou, “Speech Enhancement for Noise-Robust Speech Synthesis using Wasserstein GAN”,

INTERSPEECH, 2019.

• M. Shifas PV, N. Adiga, V. Tsiaras, Y. Stylianou, “A non-causal FFTNet architecture for speech enhancement”, INTERSPEECH, 2019.
• Y. Pantazis, D. Paul, M. Fasoulakis, Y. Stylianou, “Training Generative Adversarial Networks with Weights”, EUSIPCO 2019.
• N. Adiga, V. Tsiaras, and Y. Stylianou, “On the use of WaveNet as a Statistical Vocoder”, IEEE ICASSP, 2018.
• M. Shifas PV, V. Tsiaras, Y. Stylianou, “Speech Intelligibility Enhancement Based on a Non-causal Wavenet-like Model”, INTERSPEECH, 2018.
• A. Sfakianaki, G. P. Kafentzis, “Assessing voice features of Greek speakers with hearing loss”, 1st Conference on Interdisciplinary Approaches

to Linguistic Theory, Greece, 2017.

• G. P. Kafentzis, Y. Stylianou, “High-Resolution Sinusoidal Modeling of Unvoiced Speech”, ICASSP, China, 2016.
• A. Koutrouvelis, G. P. Kafentzis, N. Gaubitch, R. Heusdens, “High-Resolution Voiced/Unvoiced Detection and Glottal Closure/Opening Instant

Estimation of Speech”, IEEE/ACM Transactions on Audio, Speech, and Language Processing, vol. 24 (2), 2016.

• M. Caetano, G. P. Kafentzis, A. Mouchtaris, Y. Stylianou, “Full-Band Quasi-Harmonic Analysis and Synthesis of Musical Instrument Sounds

with Adaptive Sinusoids”, Applied Sciences, Special Issue on Audio Signal Processing, vol. 6 (127), 2016.

• M. Koutsogiannaki, Y. Stylianou , “Modulation Enhancement of Temporal Envelopes for Increasing Speech Intelligibility in Noise”,

INTERSPEECH, 2016.

• M. Caetano, G. P. Kafentzis, A. Mouchtaris, “Adaptive Modeling of Nonstationary Sinusoids”, International Conference on Digital Audio

Effects, Norway, 2015.

• M. Koutsogiannaki, P. N. Petkov, Y. Stylianou, “Intelligibility enhancement of casual speech for reverberant environments inspired by clear

speech properties”, INTERSPEECH, 2015.

• …
• …

UNIVERSITY OF

C R E T E

Speech Signal Processing Laboratory

UNIVERSITY OF

C R E T E

Speech Signal Processing Laboratory

Ευχαριστώ για την προσοχή σας! Thank you for your attention 