Idiap at INTERSPEECH 2021

(August 30 - September 3, 2021, Brno, Czech Republic)

Idiap's speech group will be presenting six papers at Interspeech 2021. Our papers address a variety of research problems in automatic speech recognition, natural language processing, pathological and physiological speech processing, etc.

The conference is organised in a HYBRID FORM - the conference will happen both physically in Brno and virtually, giving the opportunity to connect to all of you that won’t be able to make it to Brno.

Accepted papers:

Late Fusion of the Available Lexicon and Raw Waveform-based Acoustic Modeling for Depression and Dementia Recognition (Esaú Villatoro-Tello, S. Pavankumar Dubagunta, Julian Fritsh, et al.)

"Speech Activity Detection Based on Multilingual Speech Recognition System." (Seyyed Saeed Sarfjoo, Srikanth Madikeri, and Petr Motlicek)

"Multitask adaptation with Lattice-Free MMI for multi-genre speech recognition of low resource languages." (Srikanth Madikeri, Petr Motlicek, Hervé Bourlard)

"Contextual Semi-Supervised Learning: An Approach To Leverage Air-Surveillance and Untranscribed ATC Data in ASR Systems" (Juan Zuluaga-Gomez, Iuliia Nigmatulina, Amrutha Prasad, Petr Motlicek, Karel Vesely, Martin Kocour, Igor Szoke)

"Boosting of contextual information in ASR for air-traffic call-sign recognition" (Martin Kocour, Karel Vesely, Alexander Blatt, Juan Zuluaga-Gomez, Igor Szoke, Jan Cernocky, Dietrich Klakow, Petr Motlicek)

Handling acoustic variation in dysarthric speech recognition systems through model combination (Enno Hermann and Mathew Magimai Doss)

Comparing CTC and LFMMI for out-of-domain adaptation of wav2vec 2.0 acoustic model (Apoorv Vyas, Srikanth Madikeri, Hervé Bourlard)

"Multi-task Neural Network for Robust Multiple Speaker Embedding Extraction" (Weipeng He, Petr Motlicek and Jean-Marc Odobez)

"Modeling Dialectal Variation for Swiss German Automatic Speech Recognition" (Abbas Khosravani, Philip N. Garner, Alexandros Lazaridis)

Robust Command Recognition for Lithuanian Air Traffic Control Tower Utterances (O. Ohneiser, S. Sarfjoo, P. Motlicek, et al.)

"On Modeling Glottal Source Information for Phonation Assessment in Parkinson’s Disease" J. C. Vasquez-Correa, J. Fritsch, J. R. Orozco-Arroyave, E. Nöth, M. Magimai-Doss

"Identification of F1 and F2 in speech using modified zero frequency filtering" RaviShankar Prasad and M. Magimai-Doss

"Improving Emotional TTS with an Emotion Intensity input from Unsupervised Extraction" (Bastian Schnell, Philip N. Garner)


Show & Tell

"ROXANNE Research Platform: Automate criminal investigations", Maël Fabien, Shantipriya Parida, Petr Motlicek, Dawei Zhu, Aravind Krishnan, Hoang H.Nguyen

Idiap participates on organization of a special session on Automatic Speech Recognition in Air-Traffic Management in participation with DLR and Honeywell.