South African News24 Adds Localized Synthetic Voice to Stories
South African news provider Media24 has debuted a synthetic voice to read stories on its News24.com website with a local accent. The audio AI, designed with British voice tech developer SpeechKit, will read the articles in English with a South African accent and vocabulary. The feature serves to better brand the news service by sounding like a local and pronouncing South African names and expressions more correctly than generic text-to-speech engines.
Local News, Local Voice
Plenty of news websites offer an audio version of articles, often read in a somewhat robotic tone in a generic American accent. Though there have been plenty of improvements in cadence and vocabulary, mispronounced words and awkwardly emphasized syllables are an expected hazard of the experience. Add in accents and loan words from other languages and the baseline version may not pass muster with most listeners. South African English has a unique accent and is only one of 11 official languages of the country, meaning local names, places, and figures of speech are borrowed from tongues spanning Afrikaans to Zulu. SpeechKit worked with Media24 to make up for the gaps in how generic text-to-speech features operate in that context. The audio service is available to anyone who signs up for a digital subscription package from News24.com.
“We decided to work with SpeechKit and develop a custom synthetic voice to deliver our subscribers the value they deserved. Existing voices struggled to handle pronunciations unique to South African accents and it was imperative that the voice resonated with our readers,” News24.com deputy editor Kelly Anderson said in a statement. “We’re thrilled with the result. The voice handles local names, towns, and places better than anything we’ve heard before. It’s much more engaging to listen to a voice that sounds like our brand.”
Branding Speech
Customizing voice AI for branding and specialized services is becoming more popular, especially as it gets easier and cheaper to make high -quality synthetic voices. That interest has fueled companies like VocaliD, which offers synthetic voices for call centers and voice apps, as well as .voice cloning startups like Replica Studios and Resemble AI, who provide home tools for generating an artificial voice. For companies building voice AI on larger platforms, Amazon and Google are both expanding their global enterprise voice AI customization. Amazon’s Brand Voice feature works with companies to make unique voices for Alexa skills, including KFC Canada’s Colonel Sanders-esque voice and the National Australia Bank’s professional banker with an Australian accent. At the moment, Google’s newer platform for custom voices is still limited to American English but is expected to add more options soon. Branding the voices reading the news on websites may soon end up as just another kind of brand or logo.
“Bringing local accents to audio articles will help take this emerging format to a new level,” said SpeechKit CEO Patrick O’Flaherty in a statement. “Our focus at SpeechKit is building seamless and engaging listening experiences. By creating a voice that reflects the local audience we’re already seeing improvements in the listening data.”
Follow @voicebotai Follow @erichschwartz
Amazon is Helping Brands Create Custom Alexa Voices Starting with KFC
Cerence Debuts Tool for Customizing Your Car’s Voice Assistant