Text-to-Speech (TTS) is the conversion of digital text into a computer synthesised spoken message. Originally developed to help people with a visual impairment, computers could ‘read aloud’ any text that was typed.
Synthesised speech is now widely used in technology, business, government, academic and disability applications – examples include Sat Nav systems telling you where to go; talking smartphones telling you about the weather; or low-vision software reading the contents of your screen to you. For people who cannot speak, computer generated speech can also become their way of communicating.
The uses are many and varied. Anybody using computer generated speech for any purpose, will now be able to have messages produced by this affordable software, in local South African languages, including English with a South African accent. No longer will you need to spend a lot of money and time on human recordings of hundreds of messages, or rely on internationally accented speech synthesis. Local really is lekker.
Speech synthesis is usually created by collecting a huge database of recorded speech in a certain language. Narrators would read from a variety of texts, in order to have a full and varied collection of sounds, aiming to include every possible sound from that language in the database.
Those recordings are then ‘sliced’ up into tiny sound pieces, which represent vowels, phonemes, consonants etc. and the way they vary in different contexts. A sophisticated analysis then takes place, matching those sounds to written words. Once a voice is created, it knows how to pronounce the words and sounds from that particular language, and can read text aloud in a natural sounding way.
When computers, smartphones, answering systems etc. speak to us, they are simply using a synthesised voice to read a portion of text aloud.
In order to use the new SA Voices on a Windows* system, you would typically install the voice, go to your software’s Speech Settings, and select the SA Voice you want for your text-to-speech output.
Funded and developed the CSIR (Council for Scientific & Industrial Research, a ‘world-class research and development organisation established through an Act of Parliament in 1945, which undertakes multidisciplinary research and technological innovation that contributes to the improved quality of life of South Africans’, this is a project we can be really proud of.
These long awaited voices are set to revolutionise the South African Text-to-Speech arena. Finally, speakers of Afrikaans, isiZulu, isiXhosa, English and Sepedi (Northern Sotho) can have their devices – screen readers, spell checkers, automated phone information systems, AAC communication devices, smart homes & educational software – all speak in their mother language, with a local accent.
*Currently compatible with Windows 7, 8 and 10 only, for use with SAPI 5 speech compatible software. Also compatible with Windows 8 and 10 Ease of Access ‘Narrator’ tool for screen reading capability.
Android and iOS versions available soon, as well as a Software Development Kit (SDK) for integrating the voices into your bespoke projects.
Development projects to integrate the voices into your applications are welcomed, and bulk license options are available. Please contact us for further information.