In most cases, speeches or presentations at an international event are required to be given in a common language (e.g. English). However, for people who are not proficient in that common language, delivering presentations fluently is very difficult. Simultaneous translation seems to be a solution, but besides its high cost, simultaneous translation undermines the nature of the presentation by substituting the real voice of the lecturer as well as his/her emotions. In this research, we propose “SmartVoice”, a presentation support system, which aims to overcome language barriers. By tracking the lip motion of the lecturer, SmartVoice controls the playback of the narration, which is a sound data prepared in advance or created automatically using a voice synthesizer. SmartVoice also controls the intonation of the sound based on the position and shape of the lecturer’s mouth. As the lecturer can talk at his/her own pace with the voice automatically following, it appears as if he/she talks in his/her own voice. In our user evaluation, we confirmed that audiences find it difficult to distinguish between the narration generated by SmartVoice and that by a real voice. We also discuss the possibility of applying SmartVoice to fields other than multi-language presentation support, such as Automated Dialogue Replacement and language study.