Text-To-Speech voices are intended to read texts naturally in a human-sounding way. Unfortunately, in spite of the many improvements TTS technology has made over the past few years, Text-To-Speech voices still have an artificial/synthetic sound, as the lack of expressiveness makes them less appealing. This normally makes listening to a text longer than a short sentence read aloud by a TTS voice quite difficult.
In 2009 we developed a new voices production technique for IVONA Text-To-Speech. The aim of this technique was to mimic the voice-talents' articulacy when sounding out words, sentences, paragraphs or even whole stories and books. Our goal was to be able to create Text-To-Speech voices that sounded exactly like the voice-talents who had lent their voices to the production of TTS, preserving their full expressiveness and all their individual aspects.
We are proud to present our first two voices created using the innovation mentioned above. Both voices speak Standard
British English (Received Pronunciation accent). We've named them IVONA Amy and IVONA Brian.
Below you can
listen to a few samples of utterances:
| Original voice talent recordings | |
| Voice talent Amy | Voice talent Brian |
|
|
|
| Voice talent Kendra | Voice talent Joey |
|
|
|
| IVONA TTS voice recordings: | |
| IVONA voice Amy | IVONA voice Brian |
|
|
|
| IVONA voice Kendrabeta | IVONA voice Joeybeta |
|
|
|
The voices above were created in partnership with Royal National Institute of Blind People (RNIB), UK. Steve Taylor, Head of Innovation at RNIB says, "IVO Software has delivered amazing quality and innovation in an area that we are particularly keen to support. We are proud to have met the team at Ivo who have worked extremely hard and thoroughly - their focus is on quality and this shows in every element of our partnership".
IVONA offers a wide range of voices, but as an additional option you can order customized voices. Your voice will be created with the IVONA Rapid Voice Development technique. This semi-automatic process uses artificial intelligence algorithms. AI algorithms and their ability to learn allow us to create voice-talent-like voices with computer precision. The additional advantage of RVD is that the process of building a new voice is fast and semiautomatic. IVONA Text-To-Speech takes advantage of many other technological innovations that make it a leading system in the TTS industry. Science 2006 IVONA has maintained a leading position in the prestigious annual scientific event Blizzard Challenge (comparison of Text-To-Speech systems).
