A method for automatically enabling closed captioning in video conferencing when a heavy accent is detected from a current speaker is provided. Language background and/or ethnicity information is received as a user preference. An acceptable accent level is determined according to the user preference. An audio signal of a speaker speaking in a language is received. A pronunciation of the speaker in the audio signal is compared with standard pronunciation for the language. An accent level of the speaker is determined, and the accent level of the speaker is compared to the acceptable accent level. If the comparison determines that the accent level of the speaker does not comply with the acceptable accent level, closed captioning in enabled for the audio signal. If the comparison determines that the accent level of the speaker complies with the acceptable accent level, closed captioning is not enabled for the audio signal.