8. Perspectives and conclusions

After more than seventy years of research and industrial development, the performance of PAR systems has improved considerably, enabling us to tackle increasingly complex fields of application. The most advanced work to date concerns telephone dialogue systems, recognition of spontaneous speech, and transcription of radio and TV broadcasts. The performances obtained depend very much on the type of task considered (size and difficulty of the vocabulary, speakers, recording conditions).

From the outset, automatic speech processing has been highly dependent on technological developments. Initially based purely on analog electronics, ASR systems were later implemented on computers. Advances in microelectronics have made it possible to miniaturize and implement complex ASR systems in software form or on a chip, enabling them to be used in a wide variety of sectors, particularly...