Start typing and press enter...
It always begins the same way. A speaker records over several days a large corpus of very different texts, under the close scrutiny of our linguists. These recordings become the raw material necessary to create a synthetic voice and must cover all the possible sounds of a language. The recordings will then be sliced, tagged and organized into a voice database which, once integrated into the text-to-speech engine, will be able to transform text into speech, in real time.
But Manon’s voice is on a grander scale. We have made additional recordings, based on an enriched corpus of text. The voice database is therefore more comprehensive, allowing the text’s sense and meaning to be even better conveyed.
Natural, intelligible, fluid, rapid: the voice expresses perfectly the true sense of content being read, through intonation and depth that capture your imagination and hold your attention. Thanks to the sizeable database the voice relies on to vocalize content in real time, Manon is particularly suited to reading long texts. It’s the database that has allowed this leap forward in quality, without impacting on the speed of the vocalization process. You can judge for yourself with the voice samples at the bottom of the news or by clicking here.
‘We are very proud of Manon. Progress achieved by our Research team in improving all the processing sequence is indeed impressive. Manon has just started to be deployed and feedback from users testifies to this leap in quality. It is very thrilling for all of us at Acapela. The advances made with Manon strengthen and reward our resolve to create ever more authentic voices .’comments Lars-Erik Larsson, CEO of Acapela Group.
Manon has already been tested by the AVH (Association Valentin Haüy), the French Association for blind and visually impaired users, Acapela’s partner for many years. This partnership has been particularly close for a year now, since the very successful vocalization of audio books in AVH’s digital library ‘Éole’ using Acapela text-to-speech.
The AVH is the first to deploy Manon. Eole’s users will be able to access an audio version of the new book season’s titles in Manon’s voice.
‘Manon represents a breakthrough in the quality of text-to-speech. She pronounces specific words better than Alice – the voice that we were using until now – and the reading is more fluid. Users that have tested Manon agree that Manon is perceptibly more intelligible than Alice and some of them were really surprised by the high level of quality attained’ says Fernando Pinta Da Silva, AVH Project Manager.
Manon will soon be available for infovox4 and will be deployed shortly in all Acapela products.