Recording, analysis, statistical modeling and synthesis of singing birds

Project Type: Master/Diploma Thesis
Student: Lorenz Gutscher

Abstract

The aim of this project is to synthesize a bird specimen through training with Hidden Markov Models (HMM). Therefor research concerning sound generation, feature extraction as well as annotation has to be done. After further investigation about the sound structure and complexity of different bird species one specimen will be selected for synthesis and models will be evaluated.

Tasks:

  • Literature research on bird song structure and feature extraction
  • Feature extraction
  • HMM training
  • Evaluation

References:

[1] J. Bonada, R. Lachlan, M. Blaauw, “Bird Song Synthesis Based on Hidden Markov Models,” September 2016

[2] C. ten Cate and K. Okanoya, “Revisiting the syntactic abilities of non-human animals: Natural vocalizations and artificial grammar learning,” Philosophical Transactions of The Royal Society B, vol. 367, pp. 1984–1994, 2012.

[3] J. J. Odell, “The use of context in large vocabulary speech recognition,” in PhD Thesis, Queens’ College, University of Cambridge, Cambridge, U.K., 1995