Cluster Self-organization of Known and Unknown Environmental Sounds Using Recurrent Neural Network
Our goal is to develop a system that is able to learn and classify environmental sounds for robots working in the real world. In the real world, two main restrictions pertain in learning. First, the system has to learn using only a small amount of data in a limited time because of hardware restrictions. Second, it has to adapt to unknown data since it is virtually impossible to collect samples of all environmental sounds. We used a neuro-dynamical model to build a prediction and classification system which can self-organize sound classes into parameters by learning samples. The proposed system searches space of parameters for classifying. In the experiment, we evaluated the accuracy of classification for known and unknown sound classes.
KeywordsEnvironmental Sounds Prediction Classification Neuro-dynamical Model
Unable to display preview. Download preview PDF.
- 1.Ashiya, T., Nakagawa, M.: A Proposal of a Recognition System for the Species of Birds Receiving Birdcalls: An Application of Recognition Systems for Environmental Sound. IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences 76, 1858–1860 (1993)Google Scholar
- 4.Miki, K., Nishimura, T., Nakamura, S., Shikano, K.: Environmental Sound Discrimination Based on Hidden Markov Model. Information Processing Society of Japan SIG Notes 99, 79–84 (1999)Google Scholar
- 5.Ogata, T., Ohba, H., Komatani, K., Tani, J., Okuno, H.G.: Extracting Multimodal Dynamics of Objects Using RNNPB. Journal of Robotics and Mechatronics, Special Issue on Human Modeling in Robotics 17, 681–688 (2005)Google Scholar
- 7.Yamashita, Y., Tani, J.: Emergence of Functional Hierarchy in a Multiple Timescale Neural Network Model: A Humanoid Robot Experiment. PLoS Comput Biol. 4, e1000220 (2008)Google Scholar
- 8.Real World Computing Partnership, RWCP Sound Scene Database, http://tosa.mri.co.jp/sounddb/indexe.htm
- 9.Yamakawa, N., Kitahara, T., Takahashi, T., Komatani, K., Ogata, T., Okuno, H.G.: Effects of modelling within- and between-frame temporal variations in power spectra on non-verbal sound recognition. In: Proc. 2010 International Conference on Spoken Language Processing, Makuhari, pp. 2342–2345 (2010)Google Scholar