2022 Fiscal Year Annual Research Report
PSYPHON: Psychoacoustic features for Phonation prediction
| Project/Area Number |
20K11956
|
|---|
| Research Institution | The University of Aizu |
|---|
Principal Investigator |
Julian Villegas 会津大学, コンピュータ理工学部, 上級准教授 (50706281)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
李 勝勲 国際基督教大学, 教養学部, 上級准教授 (20770134)
MARKOV K 会津大学, コンピュータ理工学部, 教授 (80394998)
|
|---|
| Project Period (FY) |
2020-04-01 – 2023-03-31
|
| Keywords | Phonation prediction / Psychoacoustics / Machine Learning |
|---|
| Outline of Annual Research Achievements |
We conducted an extensive literature review on languages that use phonation as a contrastive cue. This review led us to perform field recordings of Zapotec and Mixe languages in Oaxaca, Mexico. We annotated and classified their phonation. Additionally, we conducted several subjective experiments primarily focusing on the effect of linguistic background on the classification of creaky words. We conducted several simulations to fine-tune the hyperparameters of the machine learning algorithms used for predicting psychoacoustic roughness. We published these results in several papers. In relation to this, we discovered that phonation types were positively associated with psychoacoustic features: falsetto with pitch, whisper with sharpness, and creakiness with loudness and roughness
|
