This is a bit of an oversimplification. Generally, they would use the laughter from the actual audience in attendance. The stands were mic’d but the nature of filming anything is that it will often take multiple takes. Ideally, you get a perfect performance and response on the first take, but that’s not reality. Maybe you got a great laugh, but Jerry clinked a glass loudly over Jason’s line. So they cut and reset, Jerry does the joke again and there’s no mistakes, but the audience response is more muted because they just heard that joke.
The solution here is pretty obvious: grab the laugh from the first take and dub that over the performance from the second take. Technically, you’re misleading the audience at home because that laughter came from a different take, but it would also be misleading to show the home audience the tenth take and you hear the audience murmur awkwardly as if they hated it, when that’s just the response you’ll get from an audience ten takes deep into hearing the same joke.
There’s even the reverse case, where maybe some audience audio just isn’t usable. Nobody notices it on the day, but there was one take you got perfectly the first time, but in editing you hear some guy sneezing loudly while the rest of the crowd is giggling. You could just lose that scene or mute the audience for it, or crossfade into some similar audio you got from the previous scene, or whatever. Other times, your actors might continue a scene but the audience laughs over the next couple of lines, so you fade the crowd. In this way, the audience response is only as fake as the show itself is. Maybe Julia gave a funnier line read in take 3 but Jason hit a run on take 5, so you edit those together, making the best of the stuff you got on the day. Sometimes it was necessary to do the same for the laughs.
It was always preferable to get the real audience response to the actual current take, because if Michael does some physical bit to play off the crowd, you should hear them respond at the appropriate time, even in the middle of a longer laugh. But sometimes the pure documentary fact of what happened in the take that made it to air just isn’t the best version of the show. Ultimately, it’s not a scheme to trick people into thinking the audience responded differently. If anything, a joke that the audience didn’t respond to would get changed on-set rather than fixed in editing. You’d huddle with the writers and go “They don’t like this, what else have you got?” Then you’d feed your actors the new lines and see if they got a better reaction.
tl;dr: Crowd sound in any sitcom that is filmed before a live studio audience is mostly genuine.
For a post-script, even pre-taped outdoor scenes and stuff would be shown to the audience on large monitors so that a) they could follow the story and b) so their reactions could be recorded in the same session, with the same crowd, including the same guy with the staccato laugh so everything sounds consistent across the entire episode.
This is a bit of an oversimplification. Generally, they would use the laughter from the actual audience in attendance. The stands were mic’d but the nature of filming anything is that it will often take multiple takes. Ideally, you get a perfect performance and response on the first take, but that’s not reality. Maybe you got a great laugh, but Jerry clinked a glass loudly over Jason’s line. So they cut and reset, Jerry does the joke again and there’s no mistakes, but the audience response is more muted because they just heard that joke.
The solution here is pretty obvious: grab the laugh from the first take and dub that over the performance from the second take. Technically, you’re misleading the audience at home because that laughter came from a different take, but it would also be misleading to show the home audience the tenth take and you hear the audience murmur awkwardly as if they hated it, when that’s just the response you’ll get from an audience ten takes deep into hearing the same joke.
There’s even the reverse case, where maybe some audience audio just isn’t usable. Nobody notices it on the day, but there was one take you got perfectly the first time, but in editing you hear some guy sneezing loudly while the rest of the crowd is giggling. You could just lose that scene or mute the audience for it, or crossfade into some similar audio you got from the previous scene, or whatever. Other times, your actors might continue a scene but the audience laughs over the next couple of lines, so you fade the crowd. In this way, the audience response is only as fake as the show itself is. Maybe Julia gave a funnier line read in take 3 but Jason hit a run on take 5, so you edit those together, making the best of the stuff you got on the day. Sometimes it was necessary to do the same for the laughs.
It was always preferable to get the real audience response to the actual current take, because if Michael does some physical bit to play off the crowd, you should hear them respond at the appropriate time, even in the middle of a longer laugh. But sometimes the pure documentary fact of what happened in the take that made it to air just isn’t the best version of the show. Ultimately, it’s not a scheme to trick people into thinking the audience responded differently. If anything, a joke that the audience didn’t respond to would get changed on-set rather than fixed in editing. You’d huddle with the writers and go “They don’t like this, what else have you got?” Then you’d feed your actors the new lines and see if they got a better reaction.
tl;dr: Crowd sound in any sitcom that is filmed before a live studio audience is mostly genuine.
For a post-script, even pre-taped outdoor scenes and stuff would be shown to the audience on large monitors so that a) they could follow the story and b) so their reactions could be recorded in the same session, with the same crowd, including the same guy with the staccato laugh so everything sounds consistent across the entire episode.
Sorry this is so long.
Long is okay, because it’s also the most accurate answer
This is a great answer.
Classic Jerry