Tecnologías de la Voz
El objetivo principal de este curso es proporcionar los conceptos y
herramientas básicas utilizadas en el procesamiento de la señal de voz.
Se presentan los conceptos básicos necesarios de acústica y se estudia con
detalle los procesos de generacón y percepción del habla y las técnicas básicas
de procesado digital de señal utilizadas en el análisis de la señal de voz.
Desde el curso 2006/07 este curso se oferta en inglés
Course Agenda 2008-2009
Part I: Fundamental Theory-
0. Fundamentals of Acoustics
- Sound and Noise
- Sound and Noise Level Measure
- Sound Propagation
- Harmonic Plane Waves
- Pipes and Cavities: Acoustic Circuits
- Anatomy and Physiology of the Speech Organs
- Acoustic Phonetics
- Acoustic Theory of Speech Production
- Anatomy and Physiology of the Ear
- Sound Perception
- Short-Time Speech Analysis
- Time-Domain Parameters
- Short-Time Fourier Analysis
- Linear Predictive Coding Analysis
- Cepstral Analysis
- Speech enhancement: Additive and Convolutive noise
- Perceptually motivated representations
- Speech Coders Attributes
- Frequency Domain Coders
- Analysis by Synthesis Coders
Part II: Speech Technologies
-
5. Pattern Recognition
- Bayes´ Decision Theory
- Classifiers
- Feature Extraction and Selection
- Why is it so difficult?
- A Basic Pattern Recognition Approach
- Statistical Speech Recognition
- Acoustic Modelling: Hidden Markov Models
- Language Modelling
- Basic Search Algorithms
- Voice Portal Development: VoiceXML
Part III: Laboratory Work
- The speech signal, acoustic-phonetic features: Generation 4h
- Short time analysis, pitch and formant estimation 4h
- Voice portals: VoiceXML 2h
- Speech recognition engine: command & control application 4h