Acoustic analysis and computerized reconstruction of speech in laryngectomised individuals

Abstract

In laryngectomised individuals, rehabilitation options include esophageal speech, tracheoesophageal puncture (TEP), and electrolarynx devices enabling patients to communicate. These do not generate natural sounding speech; at best they are monotonous or robotised. We present a non-surgical, non- invasive alternative which combines whisper signal analysis with direct pitch insertion and formant enhancement, to ‘reconstruct’ missing speech. This pilot study assesses the acoustic features of laryngectomees speech and required enhancement for natural speech regeneration.