Description: Expert in Speech Recognition, Speech Synthesis and Speech
Edmilson S. Morais
The portuguese version of this page is been prepared.
(Additional information can be found on my homepage.)
State University of Campinas - UNICAMP - SP - Brazil
+ PHD in Eletrical Engenieering (Not completed, yet) August 1997 -
August 1998 (Not completed)
+ Master Degree in Eletrical Engenieering January 1996 - December
University of Brasília - UnB - DF - Brazil
+ Undergraduate in Eletrical Engenieering January 1991 - December
CSTR (Centre for Speech Technology Research) / University of Edinburgh
Edinburgh. Scotland, UK January 2000 - July 2000
+ Visiting Researcher : Prosodic Speech Modification Using Prototype
Waveform Interpolation (A Time / Frequence Approach)
IBM - International Bussines Machine
Seville, Spain (European Speech Recognition Group) September 1998 -
+ Project Supervisor : Acoustic and Language modeling for the
Brazilian Portuguese IBM ViaVoice System. These project was
developed by team composed by : Eletrical Engineers, Computer
Engineers, Telecommunication Engineers and Linguistics.
Z-Tecnology in Communications July 1994 - January 1995
Brasília - DF - Brazil
+ Intern : Development of application for MODEM based on C++ for
Z-Tecnology in Communications May 1992 - August 1993
Brasília - DF - Brazil
+ Intern : Development of software in C, Assembly for PC and
Assembly for TMS320C51 from Texas Instruments.
GPDF - Group of Speech Digital Signal Processing of UNICAMP January
1996 - September 1998
Campinas - SP - Brazil
+ Master Degree : Application of adaptative algorithm LMS ("Least
Mean Square") and RLS ("Recursive Least Square") for speech signal
analysis. Applications of the FIR-MLP ("Finite Impulse Response -
Multilayer Perceptron") for phones analysis and phones
recognition. Analysis and Development of all modules of a Hybrid
Continuous Speech Recognition System based on ANN (Artificial
Neural Network) and HMM (Hidden Markov Model). Study and
development of algorithm for TTS-Sythesis
(Text-To-Speech-Synthesis) based on PSOLA, TFI (Time Frequence
Interpolation) and Harmonic Decomposition. Obs : During all my
work in the GPDF-UNICAMP I mantained a strong relationship with
the LAFAPE-UNICAMP (Laboratory of Acoustic Phonetic and
Experimental Psychoacoustc Group of UNICAMP).
GPDS - Group of Digital Signal Processing of UnB August 1993 -
Campinas - SP - Brazil
+ Reserch study : Study of fundamental concepts of digital signal
processing. Research and development of techniques for speech
signal processing , such as : (1) Prosodic speech modification
using a technique based on CELP/TFI (Coding excited linear
prediction / Time frequency interpolation), (2) Low bit rate
speech coding based on a CELP/TFI system, (3) Development of a TTS
system based on the CELP/TFI technique.
+ C and C++
+ Shell scripts for Unix
+ Diferencial and Integral Calculus (Teacher assistent) March 1992 -
+ Digital Signal Processing (Teacher assistent) March 1995 - July
+ Portuguese Native Language.
+ English Proficiency Level .
+ Spanish Intermediate Level : Spoken and Written.
+ Morais, E., Taylor, P., Prototype Waverform Interpolation Applied
for Concatenative Speech Synthesis ( A Time Frequency Approach) ,
Submitted for ICSLP - International Congress of Speech and
Language Processing , Beijin, China, October 2000.
+ Morais, E., Violaro F., Prosodic Speech Modification Using TFI
(Time Frequence Interpolation), International Telecommunication
Symposium, São Paulo, SP, Brazil, June 1998.
+ Ynoguti, C., Morais, E, S. Violaro F., A comparative Study Between
a Hybrid System and a Traditional HMM System for Continuous Speech
Recognition, International Telecommuication Symposium, São Paulo,
SP, Brazil, June 1998.
+ Morais, E., Violaro F., Automatic Continuous Speech Recognition
Based on ANN-HMM Hibryd System, Master Degree Thesis, UNICAMP,
Campinas - SP, Brazil, 1997.
+ Morais, E., Violaro F., Netto M. L. A., Hidden Markov Model
Trained by a ANN and Its Application for Speech Recognition, III
CBRN - Brazilian Congress of Artificial Neural Networks,
Florianópolis - SC - Brazil, 1997.
+ Morais, E., Violaro F., Netto M. L. A., Hybrid Hidden Markov Model
/ Neural Network System Following MAP and ML Aproaches and their
Application to Time Series Recognition , III SBAI - Brazilian
Simposium of Inteligent Control , Espirito Santo - ES - Brazil,
+ Morais, E., Violaro F., Hybrid System ANN-HMM for Continuous
Speech Recognition, XV SBT - Brazilian Telecommunication
Symposium, Recife - PE - Brazil, 1997.
+ Morais, E., Silva L. M., A Comparative Study of Prototype
Transmission Rate for TFI Speech Coding, TELEMO 96 - XIV Brazilian
Telecommunication Symposium and VII Brazilian Optic and Microwave
Symposium, Curitiba - PR - Brazil, 1996.
+ Morais, E., Silva, L, M., Cobra, D, T, Q., System CELP/TFI for
Speech Coding and Its Application for Text To Speech Synthesis.,
Final Project of Undergraduation in Eletrical Engenieering,
University of Brasília, Brasília - DF - Brazil, 1995.
+ Morais, E., Cobra, D, T, Q., Prosodic Speech Modifications (Pitch
and Time Scale), Anais of the First Congress of Science and
Culture of the University of Brasília - DF - Brazil, 1995.