ScotRail is set to replace a controversial AI voice on trains, after criticism from a professional voiceover artist. Gayanne Potter's Scottish accent was used to teach station announcer "Iona", but she said it was a surprise to hear a "dreadful" robotic version of herself. ScotRail introduced the voice in May, provide by Swedish tech firm ReadSpeaker, to replace pre-recorded human announcements on some services. Transport Scotland said the rail operator now intended to introduce an alternative voice "as soon as practicable". ScotRail has not confirmed if this will be a human recording or another AI-trained voice.

Text-to-speech (TTS) technology isn't exactly new – but the way it's shaping the future certainly is. From smart speakers to voice assistants, TTS is increasingly paramount in day-to-day interactions between brands and end users, leading to enhanced brand experiences and better business outcomes. Up until recently, TTS was confined to a specific use case: voice-enablement of written content to make computers'speak' to those with visual or reading impairments. TTS technology was based on utility and a need to make screen-related content accessible. As such, synthetic speech was traditionally digital-sounding and marred by poor audio quality and speaking style.

First things first: what is TTS? TTS or Text-to-Speech technology converts text into spoken speech. If you know Siri or those handy voice GPS directions on smartphones, then congratulations! Since 1000 AD, humans have strived to create synthetic speech, but it didn't enter the mainstream until the mid 1970s – early 1980s when computer operating systems began implementing it. Walt Tetschner, leader of the group that produced DECtalk in 1983, explains that while the voice wasn't perfect, it was still natural sounding and was used by companies such as MCI and Mtel (two-way paging).