Recently, the application of automatic speech recognition has become more and more widespread, and this tendency is expected to continue. One of the reasons for this spreading is the improving accuracy of speech recognition, which is due to the speeding up of computers, and also to continuous resarch and development in the field. Another reason for the spreading is the increasing popularity of mobile devices that require different input methods from the traditional keyboard and mouse, such as touchscreens and speech recognition.
A resulting question is whether chemical applications could be improved by adding speech recognition functionality, so that we can control them using our voice. To answer the question, we must understand the difficulties of creating a chemical speech recognition system, and determine the expected results. The aim of this thesis project was to answer these questions, and to show a prototype of speech recognition, which can be added to chemical software.
This paper will present the problem and challenges behind speech recognition as applied to the chemical domain, and then it will give solutions. As part of the project, a chemical speech recognition program has been created, which mainly understands chemical names, but also simple instructions with a small modification. This was integrated, in a simple form, with a chemical drawing application called MarvinSketch for demonstration purposes. The paper will document the above, and along with evaluating the created solution, it will suggest further improvements.