About
Subscribe
  • Home
  • /
  • Innovations
  • /
  • On the phone, South Africans have an ear for local English

On the phone, South Africans have an ear for local English

Johannesburg, 19 Mar 2008

It's official. Seff Efrican English rules! It will probably come as little surprise to most South Africans to know that we prefer to communicate with our own kind when it comes to self-service applications - even if we know the voice at the other end of the line is automated.

An informal study conducted by students at North-West University, in which users were asked to compare two text-to-speech systems, found that the users far preferred to chat to a local English-speaking system called Tessa, rather than its US counterpart Samantha.

Developed by US company Nuance Communications, a supplier of speech and imaging solutions, and Intelleca, South Africa's leading provider of voice and contact centre solutions, Tessa is a South African version of Nuance's RealSpeak text-to-speech (TTS) engine that dynamically converts any input text into spoken output. This is the first application of its kind for the South African market. "The research showed the respondents not only felt more at home conversing with a 'local', but that they also understood what Tessa was saying far more easily," says Dr Etienne Barnard, of the Intelleca speech laboratory. "Of the sample surveyed, the vast majority preferred Tessa's voice for her 'naturalness' and 'understandability'. Basically, she is friendly, expressive, and highly intelligible."

Phonetically, South African English is closely related to Australian and New Zealand English and to the English of south-eastern England, in which the southern hemisphere dialects have their roots. This may explain why English-speaking South Africans have a preference for non-American accents.

Tessa (Text-to-Speech South Africa) can be used to deliver premium voice quality and natural-sounding synthesised speech for call centre and network-based speech applications. Such voices are often used with voice recognition technology, which is familiar to anyone who has used self-service applications for banking, flight information services and sport results queries.

Barnard says Tessa was developed to support pronunciation of frequently used South African names and words, with specific scripts created to cover such terms. The scripts were created from scratch, as South Africa's languages are new in Nuance's global catalogue of more than 40 text-to-speech engines.

"The study confirms that the creation of Tessa will boost adoption of text-to-speech applications in South Africa," says Intelleca CEO Mike Renzon, "From reading newspapers out loud, to directory inquiries, from SMSes being read out to stock exchange feeds, the potential uses are endless. South Africans are going to become used to hearing Tessa on their phones."

Share

Editorial contacts

Karen Heydenrych
Predictive Communications
(011) 608 1700
karen@predictive.co.za
Mike Renzon
Bytes Connect
(011) 442 4242
miker@intelleca.co.za