“The deep artificial neural networks are inspired from the hierarchical structure of the human brain. Therefore, questioning the neural networks to learn robust, generalizable representations (abstractions) just like the human brain is not absurd, but challenging. However, revisiting fundamentals can be helpful.”
(August 30 - September 3, 2021, Brno, Czech Republic)
Understanding the impact of self-supervised pretraining approaches on low resource speech recognition.
Director of the Idiap Research Institute and Professor at EPFL, Hervé Bourlard is recognized for his major contributions to neural networks for statistical speech recognition. This distinction is awarded to him jointly with his colleague and long-time friend, Prof. Nelson Morgan of the International Computer Science Institute, and the University of California at Berkeley.
Obfuscating Voice Identity
Improving Swiss German speech recognition for Swisscom TV Box Voice Assistant
Idiap's speech group will be presenting six papers at ICASSP 2021. Our papers address a variety of research problems in pathological and physiological speech processing, automatic speech recognition, and machine learning.
“To understand others implies no only to get their point, but also to understand their feelings and emotions”; Daniel Goleman, Emotional Intelligence.
PARIDA Shantipriya, Idiap Postdoctoral Researcher has been granted to be an invited speaker at the Indo-German SPARC Symposium.
3rd ROXANNE Newsletter
Alternation in respiratory system and speech production system results in changes in speech. Therefore, speech signal, which can be acquired in a non-invasive manner, could be used to predict breathing patterns. There is a growing interest in that direction, which has gained further momentum with COVID-19 situation.
Highly Automated Air Traffic Controller Workstations with Artificial Intelligence Integration
A first step towards a Julia-based speech recognition toolkit.
You’re never far from an English word in Switzerland
As speech processing expands with diversified tasks, designing novel neural network models capable of learning meaningful speech representations gain importance.
This week Idiap's speech and machine learning group presented their joint work on Fast Transformers with Clustered Attention
Prof. Esaú Villatoto Tello, visitor professor at Idiap since September 2019, joins the SARAL project to work on the design of Cross-lingual Information Retrieval tools.
Lei joined Idiap in October 2019 for an internship in the framework of the China Scholarship Council visiting scholar program. During her internship, she worked on neural network-based mappings for single-channel dereverberation and noise reduction.
Roxanne is a EU-funded project that leverages text, speech and video in real-time in order to build tools for combating organized crime.
This week Idiap's speech group will be presenting five papers at Interspeech 2020, the largest conference for automatic speech processing. Our papers address a variety of research problems: Pathological speech processing, Multilingual automatic speech recognition, Automatic Speech Recognition for Air Traffic Control management and Speaker recognition.
Organized by the KIIT University, India, this 5 days program involved Idiap speakers.
Coordinated by Idiap, the Roxanne project aims to introduce AI technologies for law enforcement agencies. A few days ago, criminal investigation TV series were among the first data sets used to demonstrate early technologies developed by the project partners.
Edition: April 2020 and September 2020
Full time professor in Mexico, Esaú Villatoro is on sabbatical year at Idiap. He aims to develop a project in the Natural Language Understanding field with different researchers from the Idiap Research Institute.
Coordinated by the Idiap Research Institute, the European project Roxanne gathers a large variety of national and international police forces, including Interpol, as well as scientists, large industry and private companies. Its aim is to create a computer programme able to help investigators to link various clues and uncover criminal network activities.
Interpol’s New Software Will Recognize Criminals by Their Voices
Following a very strict evaluation process – comprising the nomination by Idiap's management and a formal approval by the Scientific College - , Idiap is pleased to announce the promotion of three of its researchers to Senior Researchers:
LYON, France – In its final field test, the Speaker Identification Integrated Project (SiiP) successfully demonstrated the system’s innovative capabilities as language independent voice recognition system.
The work described in this thesis takes place in the context of data-driven integration of linguistic knowledge and acoustic information for the pronunciation lexicon development.
When speech processing systems are designed for use in multilingual environments, additional complexity is introduced. Identifying when language switching has occurred, predicting how cross-lingual terms will be pronounced, obtaining sufficient speech data from diverse language backgrounds: such factors all complicate the development of practical speech-oriented systems. In this talk, I will discuss our research group's experience in building speech recognition systems for the South African environment, one in which 11 official languages are recognised. I will also show how this relates to our participation in the BABEL project, a recent 5-year international collaborative project aimed at solving the spoken term detection task in under-resourced languages.
Revisiting autoencoders for speech representation learning
