The paper describes a reliable system capable to read typewritten documents and to convert them into a speech output. The basic architecture of the system is built around a commercial optical scanner connected to a Personal Computer provided with an add-on DSP card; this card represents the heart of the speech capability of the entire system; the board, entirely developed at the research centre of Alcatel FACE, allows several speech processing features, as connected words speech recognition, speech compression and Italian text to speech conversion. The text to speech conversion, also developed at our laboratory, is based on the segment concatenation approach where the basic segments are diphones and triphones; the current size of the segment database is in the range of 400; both male and female voices are available by means of two separate set of segments and if required the type of voice can be changed in real time.
Bibliographic reference. Abbattista, Giuseppe / Riccio, Antonello / Mumolo, Enzo (1991): "Automatic document reader with speech output capabilities", In EUROSPEECH-1991, 321-324.