A Review on Voice Conversion Approaches for Voice Enhancement of Voice Artists


Pasindu Ekanayake and K.B.N. Lakmali


Voice Casting, Voice Acting, Voice Conversion, Source to Target

Issue Date:

18th February 2022


The entertainment industry has long valued voice acting, and it is critical to the success of any product. When casting, inexperienced and even experienced casting directors find it difficult to select the most suited voice actor for the role and indie producers find it expensive to cast many talented voice actors at once. Hence, a voice enhancement technology that can transform a given voice to a target voice would let casting directors to concentrate on casting talented voice actors without dwell on too much about the voice of the actor. It would also assist producers in cutting costs by allowing them to cast few skilled actors and change their voices according to each character. Voice conversion is a method for preserving linguistic information while modifying non-linguistic information in a voice. Various methods, ranging from statistical modeling to more recent deep learning approaches, have been experimented throughout the years. This paper assesses how the field of voice conversion can be applied to this problem domain, as well as the best voice conversion methodologies and an analysis of existing models. This review was conducted as part of an ongoing research, and the insights will be used to develop a voice enhancing system capable of transforming a given voice into a target voice.