Here is a look at a couple ways to take background sound off and leave voice only on video.
Video and audio post-production is as much about cleaning up what you captured as it is about arranging it. Audio is especially difficult in that it is often difficult to control exactly what sounds you intend and those that are in the background. When you are recording things like dialogue, you will find that you tend to end up with a lot of background sound behind the voices that you want, and this can often play against your ability to cut together the footage in continuity editing or create a world that the audience can suspend their disbelief about. Here is a look at how to take background sound off and leave voice only on video that you have captured, though it can never happen perfectly.
If you are trying to learn how to take background sound off and leave voice only on video, you are going to find that the primary way to approach this is by isolating different tones that you want to remove in audio post-production. When you are in an audio mixing program like Soundtrack Pro you will want to set a noise print, or a general print for the tone you want to remove. This is going to be the background sound, and you will find it in gaps in the dialogue. You will then use EQ, noise reduction, or other tools to create a setting that generally removes that audio tone. You will then apply that alteration to the entire audio track, ripping out the background tones from the dialogue.
This will alter the sound of the dialogue somewhat since the tones in the background sound are also in the dialogue. As long as the background sound is fairly consistent and the dialogue is far above it then this should not be much of a problem as long as you sweeten the dialogue afterwards.
ADR, or automatic dialogue replacement, is an option when you need a much more powerful tool when figuring out how to take background sound off and leave voice only on video. What ADR tends to be used for is if the background sound or noise is making the dialogue unusable, and it is essentially where you record the dialogue again according to the video of the scene. To do this you often use an audio mixing booth along with a narration booth. The actor is positioned in front of a screen that will provide playback of the scene and they will then be asked to speak along with the scene as it plays out. This will often be one line by line many different times, trying to line it up as closely as possible. The lines are then placed in the audio timeline right where they occur so that they will be synced up to the video, and compression can be applied so that the waveform of the new audio will somewhat match the waveform for the damaged audio. This is not exactly a method for how to take background sound off and leave voice only on video, but instead an option to recreate the audio track with no background noise under the voices.
Source: author's own experience