Unformatted text preview: CSD 250 CSD 250 Speech Science
The sourcefilter theory of speech production Last time: Complex waves Last time: Complex waves Quality, complexity Fourier analysis Harmonics Spectrum Repetitive vs. nonrepetitive waves Today: Sourcefilter theory of Today: Sourcefilter theory of speech production Müller, 1848 Articulatoryacoustic Articulatoryacoustic Relationship Speech production Speech production Sourcefilter theory Sourcefilter theory
Supralaryngeal Filter Laryngeal Source Source Source Provides energy for a wave Examples: stone thrown to a river, tuning fork, falling book hitting the ground, piano or violin strings Source for speech Source for speech Laryngeal: Vocal fold openingclosing (e.g. vowels) Noise: Frication noise generated through narrow constriction (e.g. fricatives) Transient: Sudden release of vocal tract closure (e.g. stops) Multiple sources? Source: Acoustically Source: Acoustically Laryngeal Transient Noise Filter Filter What passes through? What passes through? Transfer function Lowpass filter Another Another transfer function Highpass filter Bandpass filter Bandpass Analogy to Sound: Analogy to Sound: Acoustic Filter Source + filter = output Source + filter = output
Source Filter Output Output = source + filter Output = source + filter Filter for speech Filter for speech production: SVT SVT as SVT as compound filter/resonator Resonant frequencies: Frequencies at which acoustic energy generated at source can pass through SVT filter Also called natural frequencies Vowels & formants Vowels & formants Formants: Natural frequencies of the vocal tract Vowels are distinguished based on formant frequencies Resonance Resonance Resonators Resonators Simple resonators Compound resonators
– Saxophone – Tuning fork – Guitar Valve: Source Body: Resonator String: Source Body: Resonator Resonant frequency Resonant frequency A simple vibrator has only one natural mode of vibration The frequency of the natural mode is the resonant frequency or natural frequency Resonance in speech Resonance in speech The supralaryngeal vocal tract is a compound resonator: More than one natural frequency Waves with frequencies near the natural frequencies can pass, while others are suppressed or dampened Source & filter are Source & filter are (sort of) independent Same source characteristics, different filter Different source characteristics, same filter Again, why source, filter, & Again, why source, filter, & resonance? Because they characterize how speech sounds are produced and how the acoustic output is generated, based on which speech sounds are distinguished ...
View Full Document
- Winter '10