After 4 years of work experts from the Wrocław University of Technology and Polish Unikkon Integraf created an Intelligent System of Speech Recognition and Transformation into Text - MagicScribeMedical.

This is the first solution of the kind meant for the medical industry, it recognises medical vocabulary, medicine names and medical procedures. MagicScribeMedical recognises dictated text and converts it into a written form. The text is then saved as a written document in a file form on computer disc. The system has two work modes: direct dictation of a description (e.g. during a check-up) or a transformation of previously recorded digital files (e.g. into a dictaphone or other portable digital device). The system’s capacity accounts for 160 words a minute i.e. 3 times the speed of a person typing on a computer keyboard.

The company promotes the system with a slogan "You name it, we write it". The recognition and transformation takes place in real time with 98% efficiency - ensures the vice president of Unikkon Integral, Daniela Grabowska. Ms Grabowsa adds that the system draws heavily on the so called Branch Speech Corpus. The company has access to Speech Corpuses in the field of diagnostics, among others, in RTG, USG and tomography . Corpuses for other diagnostics types are under construction and should be completed by the end of the year making thus the complete speech body for the medical industry available. The aim of the system is to streamline and accelerate patient servicing in medical centres and to improve doctors' working conditions with simultaneous labour cost reduction.

Creators of the system emphasise that MagicScribeMedical does not make mistakes. The system learns correct spelling and consequently should be able to recognise the dictated words with precision equalling 98%. Authors plan to implement the system in big medical centres. The system may shorten the time devoted to medical documentation what may result in measurable financial benefits. The Research & Developemnt Centre Unikkon Integral plans to create similar systems for use in other fields of life. At the moment prototype versions of a similar solution is available for public administration. The current stage of this project development indicates that within a year we will be able to create a system enabling users to send mobile text messages with the use of their voice. - informs Daniela Grabowska. (PAIIZ)

Source: Polish Market Online

You can discuss more about Speech Recognition and related topics in our Speech Recognition Group.