Not having a voice can be a nightmare. It affects communication and self-esteem, promotes frustration and can trigger depression. To respond to millions of people who are affected by a lack of voice – or a dysphonic voice – a group of Portuguese researchers has just started a project that aims to create an innovative system for the reconstruction of the natural voice.
Under the designation DyNaVoiceR, the project gathers together engineers, otolaryngologists and speech therapists, from the Faculty of Engeneering of the U.Porto (FEUP), from CINTESIS and from the University of Aveiro, respectively. “The objetive is to create an advanced technology aid that can convert whispered speech signals into natural speech signals”, says Aníbal Ferreira, principal investigator and professor at (FEUP), highlighting that there is no known effective system that gives a convincing solution to the problem of oral communication of patients with dysphonic voice.
That is, the project wants to create an app that is a solution for all those patients suffering from a particular type of dysphonic voice: aphonia (temporary or permanent), which can result from oncological diseases (such as thyroid or larynx cancer), neurological disorders (such as multiple sclerosis or Parkinson’s Disease), psychological disorders (such as anxiety), among other causes.
“The voice is produced by the vibration of the air that is expelled from the lungs by the diaphragm and which passes through the vocal folds and is later modified by the tongue, palate and lips. The emission of a healthy voice is called euphonia, while a ‘sick’ voice, that is, with some of its characteristics being altered, is called dysphonia”, explains Jorge Spratley, otorhinolaryngologist and researcher at CINTESIS, who is responsible for the clinical area of the project.
People who suffer from chronic aphonia, speaking in whispers or with much hoarseness, see their communication capacity greatly diminished and are conditioned in their natural interaction, since the dysphonic voice transmits neither emotions nor the identity of the speaker. The system to be developed should serve as a “modified but intelligent amplifier,” which not only projects the voice by highlighting the linguistic message, but also corrects it by returning its natural sonority and even the speaker’s sound signature.
Technically, the voice can be decomposed and reconstructed to correct the intended aspects, including insertion of missing components. In this sense, it will be up to CINTESIS to collect samples of the voice in the place and at which time it is generated by the vocal cords. “It’s like getting the water directly from the spring, before passing through any creeks or pollution that alter it,” explains Jorge Spratley. For this, the researchers will draw on a sample of healthy volunteer speakers. The non-invasive exam will be held at the University Hospital Center of São João and will help collect this individual “vocal imprint.”
This information will then be worked by the engineering experts who will be in charge of developing techniques and tools for the precise synthesis and control of the different components of the voice signal, fundamental for the reconstruction of natural voice from the voice of the phonograph. Very briefly, it is intended to “preserve and highlight linguistic information, transmit distinctive elements of an individual’s voice, and improve vocal projection,” explain the researchers.
DyNaVoiceR follows other projects in the area of spoken and sung voice, such as the ARTTS project, for example, also led by Aníbal Ferreira.
Funded by the Foundation for Science and Technology (FCT), the project also has the participation of the Faculty of Medicine of the University of Porto (FMUP), the Institute for Systems and Computer Engineering, Technology and Science (INESC TEC) and the Institute of Electronics and Informatics Engineering of Aveiro (IEETA).