Abstract :- People always try to produce effortless and unobstructed speech in order to communicate. This type of speech is referred as casual speech that are produced under casual or typical circumstances when no special speaking effort is made. Casual speech is mostly simple and less articulated speech. Whereas, clear speech is hyper-articulation, with increased effort on the part of the speaker to slow down and enunciate. This clear speaking adjustment is proven to increase the intelligibility of casual speech. A speaker will deliberately change his/her speaking style in order to communicate with the listener that faces a communication barrier. For example, the target listener could be either hearing-impaired or a non-native listener. Speakers adjust their speech to achieve the maximum clarity with the minimum speech effort. Transforming casual speech to clear speech is, therefore, important but also quite challenging. Our work exactly focuses on overcoming this and try to produce well articulated speech. We use machine learning techniques to recognize the casual speech and text-to-speech synthesizer that is trying to generated well articulated clear speech.
*ENRICH has received funding from the EU H2020 research and innovation programme under the MSCA GA 675324