Piero Cosi

Signal Processing and Speech Analysis, Auditory Modeling, Automatic Speech Recognition, Neural Networks, Emotional Text-To-Speech Synthesis, Emotional AudioVisual Synthesis, Facial Animation, Robot Interaction.

Scientific Activity

Basic and applied research on experimental phonetics and speech technology. The scientific activity, mainly concerned with Speech Technology, is focused on speech signal processing, speech analysis, acoustics, psychoacoustics, auditory modeling, text-to-speech synthesis, automatic speech recognition, facial animation and speech processing for robotics.

In particular, it was focused on the development of a neuro-physiological speech processing model and of an automatic speech segmentation and labelling system (SLAM), on the implementation of the Italian version of the software named CSLU Speech Toolkit, CSLR FLT, SONIC, SPHINX, JULIUS, BAVIECA, KALDI, FESTIVAL, MBROLA, MaryTTS, dealing with Automatic Speech Recognition and Text-to-Speech Synthesis. As for facial animation, various activities are focused on the development of BALDINI, GRETA and LUCIA, three Italian emotive and expressive talking heads. As for Robotics, the activities are focused on the development of ASR and TTS systems for the NAO humanoid ROBOT for the ALIZ-E European Project.

Partial effort is devoted to the maintaining of MIVOQ s.r.l. a new company, originally born as a CNR spin-off, focused on “personalized” TTS.

Signal Processing, Speech Analysis

• "SLAM": Segmentation and Labelling Automatic Module

• Auditory Models for Speech Analysis and ASR
  
  
  • Seneff's Joint Synchrony/Mean Rate Auditory Speech Processing

  • Lyon's Passive and Active Cochlea Models

  • Auditory Inversion Techniques

  • Auditory Robust Pitch Extraction and Signal Enhanchement

Automatic Speech Recognition (ASR)

• Auditory Model for ASR

• Bimodal Phonetic Recognition by Neural Networks

• Italian ASR
  
  
  • CSLU Speech Toolkit

  • SONIC

  • SPHINX

  • JULIUS

  • BAVIECA

  • KALDI

Text-To-Speech (TTS) Synthesis

• Speech Synthesis Research - Italian Text To Speech:
  
  
  • PARLA

  • FESTIVAL

  • MBROLA

  • MaryTTS
  • MIVOQ s.r.l.

Facial Animation, Talking Heads

• Lucia

• Interface

Speech and Emotions

• "PF-STAR": Preparing Future multiSensorial inTerAction Research - IST-2001-37599

Speech Processing and Robotics

• "ALIZ-E": (Adaptive Strategies for Sustainable Long-Term Social Interaction, n. 248116, FP7-ICT-2009-4)