Freesound Forums

I was watching a TV show with a laugh track in it and thought: "Hey, I've never heard anyone DIY their own laugh track before." So naturally, I'm now going to do it.

I can do a decent number of voices, but this would work out even better if I got a bunch of other people in on it. So, here's my idea for how to do it:

- Get in a good mood or watch something funny (ideally with headphones)
- Record your genuine laughter for a while
- Upload it to freesound under CC0 Public Domain license

I'll cut and edit everything so that the timing of all our laughter sounds natural, as if we're one crowd watching one thing.

So, how about that? Anyone else interested?

strangehorizon wrote:
... I've never heard anyone DIY their own laugh track before ...

There's a reason for that: It's going to be more difficult to create than the real thing.

You'd need to neutralize the EQ of each of the laughter submissions, (imposed by the different mics, room & gear). Then put them in a virtual TV-studio space.

[ Looks like a job for AI which has learned from actual studio laugh-tracks ].

I think we can gate or de-reverb 'em and then run them through an impulse response to give them a common filtering. The drier the final signal can be, the more convincing it should be. I never really hear reverb in any studio laugh tracks. Even when someone laughs at the top of their lungs, there's only a few ms of anything like reflections.

I suppose I could download a bunch of CC0 freesound laughter to make a proof-of-concept. Yeah, I'll do that and try the EchoThief impulses. A lot of those sound really natural so they should work in small doses.

Well, freesound went under maintenance right after I wrote that last post, so I was only able to work with my own laughter samples. I was able to make a convincing crowd of 6 people. I'll try incorporating others' voices after I have a meal here.

Right, so, I've made decent progress. I was using Acon DeVerberate to de-reverb some freesound laughter, but Bertom Denoiser did most of the heavy lifting.

These are just panned into position, denoised, de-reverb'd (when needed) and then put through a master bus denoiser, then finally Boogex with the EchoThief "Cranbrook Art Museum" impulse loaded, 10% wet level. Boogex's reverb is turned off so it's just filtering.

Right now I'm just running a TRS cable straight from my PC soundcard to my speakers' aux in, so I might not hear the ultra fine details. I need a new interface or preamp. But, it sounds like these sounds gel together pretty well to me.

v1 - Zeph
v1 - Zeph + various Freesounders

I think a lot of what makes this convincing is in the timing and arrangement. I don't play every voice every time.