000000717 001__ 717
000000717 005__ 20251218164303.0
000000717 0247_ $$2DOI$$a10.6083/M4ZG6Q7F
000000717 037__ $$aETD
000000717 245__ $$aPerceptual cost function for cross-fading based concatenation
000000717 260__ $$bOregon Health and Science University
000000717 269__ $$a2012
000000717 336__ $$aThesis
000000717 502__ $$bM.S.
000000717 502__ $$gComputer Science & Electrical Engineering (sunsetting)
000000717 520__ $$aConcatenative synthesis, the dominant Text-to-Speech (TTS) method, often produces audible discontinuities due to mismatched phonemic and prosodic contexts. Previous linear cross-fading approaches improved smoothness but generated unnatural formant trajectories. This thesis proposes a unit-dependent, parameterized cross-fading algorithm guided by a perceptual cost function predicting speech quality from acoustic distance measures. Using a custom corpus and perceptual experiments, we show that output quality depends on formant trajectory shape across the vowel and correlates with both absolute distance and its derivative. Results demonstrate feasibility of perceptual cost-based optimization for natural-sounding TTS, advancing speech synthesis beyond traditional concatenation techniques.
000000717 540__ $$fCC BY
000000717 542__ $$fIn copyright - single owner
000000717 650__ $$aCommunication Aids for Disabled$$016936
000000717 650__ $$aAuditory Perception$$015222
000000717 6531_ $$aspeech synthesis
000000717 691__ $$aSchool of Medicine$$041369
000000717 692__ $$aDepartment of Computer Science and Electrical Engineering$$041404
000000717 7001_ $$aMiao, Qi$$uOregon Health and Science University$$041354
000000717 7201_ $$avan Santen, Jan$$uOregon Health and Science University$$041354$$7Personal$$eAdvisor
000000717 8564_ $$978688f63-1ada-4a5b-b372-957282babb3a$$s239197$$uhttps://digitalcollections.ohsu.edu/record/717/files/720_etd.pdf$$ePublic$$22da545c12d69dac9eed983ccdf84f70c$$31
000000717 905__ $$a/rest/prod/cr/56/n0/99/cr56n0995
000000717 909CO $$ooai:digitalcollections.ohsu.edu:717$$pstudent-work
000000717 980__ $$aTheses and Dissertations