SpeeD Lab is dedicated to scientific research

The Speech and Dialogue Laboratory is currently concerned with scientific research in all areas of Spoken Language Technology, Natural Language Processing, Speech Emotions Recognition, Machine Learning and Deep Learning, Robotics, Intelligent Robots and Human-Robot Interaction.

The academic activity is a top priority for SpeeD

SpeeD Lab is dedicated to academic teaching and supervising at bachelor, master and doctoral level within University Politehnica of Bucharest. Among others, we are teaching Microcontrollers and Microprocessors Architectures, Spoken Language Technology, Digital Signal Processing and Advanced Techniques in Digital Signal Processing.

The 11th International Conference on Speech Technology and Human-Computer Dialogue

SpeD 2021 is a conference and international forum which reflects the latest tendencies in spoken language technology and human-computer dialogue, machine learning for audio, speech, image and multimodal information processing, biometrics and security for IoT, intelligent robots and embedded systems.

SpeeD Lab benefits from a modern research & teaching infrastructure

SpeeD Lab uses several multi-functional rooms (offices, meeting rooms, teaching and research laboratories), equipped to fully accomodate academic and research activities: powerful laptops, high-performance computing systems, GPU cluster, robots, robotic arms, aerial systems, development boards.

SpeeD: an enthusiastic, passionate and professional team

SpeeD team is composed of several professors, lecturers, and PhD/Master students, but is continuously growing as more enthusiastic students are joining us. Our teaching and research activities, as well as our industry collaborations make SpeeD a dynamic and interesting place to work.

 

Home

About SpeeD

Speech and Dialogue (SpeeD) Laboratory is a teaching and research laboratory within University “Politehnica” of Bucharest. SpeeD Lab was founded in 1984 by Professor Corneliu Burileanu, who is also the leader of the laboratory ever since. SpeeD Lab’s staff is currently composed of 6 full-time academic members (2 full professors, 1 associate professor, 1 lecturer and 2 assistants), 5 researchers, more than 10 PhDs and several master students. For almost 20 years, SpeeD organizes Speech Technology and Human-Computer Dialogue Conference – SpeD, an IEEE-indexed scientific international event  once at two years, gathering academics and industry professionals from universities, government agencies and companies.

Research activity

The Speech and Dialogue Laboratory is currently involved in several research projects and is concerned with scientific research in multiple areas:

  • all areas of Spoken Language Technology, including:
    • Automatic Speech Recognition and Text-to-Speech synthesis
    • Speaker Recognition
    • Spoken Term Detection, Spoken Document Indexing/Retrieval
  • Natural Language Processing, including Language Modeling
  • Speech Emotions Recognition
  • Human-Computer Interaction
  • Digital Signal Processing
  • Machine Learning and Deep Learning
  • Embedded Systems
  • Multimodal and Multimedia Processing
  • Robotics
  • Intelligent Robots and Human-Robot Interaction

SpeeD Laboratory is one of the leading speech research groups in Romania, with many significant contributions to the development of spoken language technologies, among which the most important are:

  • the first small vocabulary isolated words recognition system for Romanian (1986);
  • an early version of a complete Voice over IP (VoIP) system (1994-1995, see the Proceedings of “ICSPAT” Conference, Boston, USA, 1995);
  • the first version of a Text-to-Speech (TTS) synthesis system in Romanian (1998);
  • the first Large vocabulary Continuous Speech Recognition (LVCSR) system for Romanian (2011);
  • a Rich Speech Transcription service for audio documents – LVCSR-ROM project (2014);
  • a natural-language, voice-controlled assistive system for intelligent buildings prototype – ANVSIB project (2017);
  • an automatic infant crying recognition system – SPLANN project (2017).

Academic activity

SpeeD Lab’s staff is dedicated to academic teaching and scientific research at bachelor, master and doctoral level in various fields at the confluence of electronics, signal processing and information technology. SpeeD Lab’s staff is teaching several courses within ETTI Faculty, among which the most important are: Microcontrollers and Microprocessors Architectures, Spoken Language Technology, Digital Signal Processing and Advanced Techniques in Digital Signal Processing. Furthermore, from 2011 the group coordinates the two-years BIOSINF master programme (“Multimedia Technologies in Biometrics and Information Security Applications”).

Infrastructure

SpeeD Laboratory is located in two places: theFaculty of Electronics, Telecommunications and Information Technology (ETTI) and CAMPUS Research Center, where it uses several multi-functional rooms (offices, meeting rooms, teaching and research laboratories). Both locations are equipped to fully accomodate academic and research activities: powerful laptops, high-performance computing systems, GPU cluster suitable for Machine and Deep Learning, development boards, robots (NAO, Kuka youbot, Jaguar), robotic arms, robocontrollers (hexapods),  aerial systems (quadcopters, SUAV), networking hardware, professional audio recording/playback devices, measurement and analysis equipments, software development tools and dedicated software applications, multimedia facilities for teaching and presentations, etc. More about the infrastructure: Robots – Autonomous and Adaptive Systems Lab and Human-Computer Dialogue Systems and Speech and Audio Forensics Lab.

Partnerships

Since its foundation, SpeeD has coordinated and was involved in numerous national and international research projects and industrial projects. Among the collaborators, there are:

SpeeD has successfully implemented over 30 private and national research grants. SpeeD group holds more than 10 Romanian issued patents and its members have authored more than 300 books, book chapters and journal/conference papers.