Related Experiment Videos
A method for generating natural-sounding speech stimuli for cognitive brain research
P Alku1, H Tiitinen, R Näätänen
1Helsinki University of Technology, Acoustics Laboratory, Finland. paavo.alku@hut.fi
Summary
Semisynthetic speech generation (SSG) creates natural-sounding speech stimuli for brain research. This method combines artificial processes with natural human vocal tract features for enhanced cognitive experiments.
Area of Science:
- Cognitive Neuroscience
- Speech Science
- Signal Processing
Background:
- Growing interest in using human voice for cognitive brain research.
- Need for high-quality, controllable speech stimuli in experimental settings.
Purpose of the Study:
- Introduce a novel method, semisynthetic speech generation (SSG), for creating speech stimuli.
- Enhance the quality and controllability of speech stimuli for cognitive research.
Main Methods:
- SSG synthesizes speech by combining artificial processes with natural human speech production.
- Estimates glottal flow from natural utterances using inverse filtering.
- Uses glottal flow to excite an artificial digital filter modeling speech's formant structure.
Main Results:
- SSG produces speech stimuli of superior natural quality compared to commercial synthesizers.
- The man-originating glottal excitation contributes significantly to the naturalness of the synthesized speech.
Conclusions:
- SSG's artificial vocal tract modeling allows adjustable formant frequencies.
- This makes SSG a suitable method for cognitive experiments utilizing speech sounds as stimuli.