Audioflow
Morphed /ay/ transition

Morphing Spectral Envelopes Using Audio Flow

Tony Ezzat, Ethan Meyers, Jim Glass, Tomaso Poggio

Center for Biological and Computational Learning (CBCL) &
Computer Science and Artificial Intelligence Lab (CSAIL)
Massachusetts Institute of Technology
Cambridge, MA

link to paper: (ps.gz) (pdf)

Abstract

We present a method for morphing between smooth spectral magnitude envelopes of speech. An important element of our method is the notion of audio flow, which is inspired by similar notions of optical flow computed between images in computer vision applications. Audio flow defines the correspondence between two smooth spectral magnitude envelopes, and encodes the formant shifting that occurs from one sound to another. We present several algorithms for the automatic computation of audio flow from a small 20 second corpus of speech. In addition, we present an algorithm for morphing smoothly between any two spectral magnitude envelopes, given the computed audio flow between them.

Results 1 - Examples of audio flow computed by our various algorithms
Results 2 - Movies of real and morphed transitions
Results 3 - Real and morphed sounds
Algorithmic Details - Some more details on the algorithms & corpus

Related Paper

Morphing Spectral Envelopes using Audio Flow Tony Ezzat, Ethan Meyers, Jim Glass, and Tomaso Poggio, to appear: Interspeech/Eurospeech 2005, Lisbon, Portugal (ps.gz) (pdf)

Last updated August 25, 2005. Send any comments or questions to Tony Ezzat