Klatt's last tapes.

Professor Stephen Hawking’s daughter, journalist and novelist Lucy, traces the development of artificial speech synthesizers over the last 200 years. Lucy visits Saarland University in Germany, where researchers Jürgen Trouvain and Fabian Brackhane have built a model of the first ever voice machine, created in the 18th century by inventor and showman Wolfgang von Kempelen. Von Kempelen was inspired by the pipe organ stop ‘vox humana’ which seemed to mimic human speech. Using the replica, Lucy recreates the only two words it could produce: ‘mama’ and ‘papa’. Von Kempelen recognised the limitations of his device, which lacked a tongue and was thus limited to rudimentary vowels. Professor Bernd Möbius, also of Saarland University, discusses the first electronic synthesiser – the VODER – invented by Homer Dudley in the 1930s. We hear recordings of its demonstration at the 1939 World’s Fair. The machine was difficult to use; its female operator is heard saying it took a year of ‘constant practice’ to become an expert. Lucy interviews a PHD student working on articulatory synthesis of women’s voices, which are harder to simulate as higher pitches sound artificial. The problem also exists for children; Lucy points out that it must be difficult for children with ‘adult’ voices to bond with others in their age group. The ultimate aim would be to create a voice that can grow with the child. Möbius introduces Dennis Klatt, whose pioneering work on text-to-speech technology allowed non experts to use voice synthesisers for the first time. In the 1980s Klatt created the DECtalk synthesiser; the electronic voice used by Stephen Hawking. Klatt’s daughter Laura Fine describes his desire to create a natural and intelligible system. Uniquely, he also worked with perception data to establish how cues relating to age and gender are interpreted by the listener. Klatt took snippets of his daughter’s voice to develop a children’s preset for DECtalk. Ironically, as a 40 year old Klatt himself developed a raspy voice as a result of thyroid cancer. Lucy then hears from Michael Cubiss, who lost the ability to speak after a sudden stroke. He uses a synthesiser equipped with eye gaze technology – a text-to-speech system operated by the eye movements of its user. Pete Donegan, who fitted the system, discusses brain computer interfaces for ‘locked-in’ patients, which he sees as the future of speech synthesis. Acclaimed Irish filmmaker and MND sufferer Simon Fitzmaurice discusses his synthesised voice – a generic American model which has nevertheless become part of his identity. The researchers at Saarland are working on an application for patients with similar conditions, allowing them to record their speech before it is lost in order to generate a personalised synthetic voice. Unnerved to hear someone else using ‘his’ voice, Fitzmaurice has been working with a company in Edinburgh to produce a bespoke version using old recordings and his father’s speech – although he is yet to decide whether or not he will adopt the new voice. The programme is interspersed throughout with Klatt’s archived recordings (preserved by the Acoustical Society of America) charting the progress of speech synthesis in the 20th century.