The invention provides an information processing apparatus, a conference system and an information processing method. In a terminal apparatus used by a speaker, sounds are inputted via a microphone to perform sound recognition processing and morphological analysis, a character string obtained as a result of the analysis is extracted using a predetermined condition, and the extracted character string is transmitted to other terminal apparatuses via a conference server apparatus. On each of the other terminal apparatuses, the extracted character string, which has been received, is displayed in a selectable manner. The selected character string is displayed in a superimposed manner on an image of shared document data. The character string converted from sounds uttered by the speaker at a conference is freely placed on the shared image, thereby effectively aiding a conference participant to make a note at the conference.