AI Weirdness

Neural net does sound effects

The BBC has published their entire archive of 16,000 sound effects, recorded over many decades and available for public use. Existing sound effects include “Wild boars having tea”, “4 batter puddings thrown”, “Several men snoring hilariously,” and “indisposed chicken,” along with lots of horses, engines, and clocks. (NOTE: as with most of the links in this article, sound will play as soon as you visit the link)

I’m not sure what the BBC intended people to use these sound effects for, but neural network enthusiasts immediately recognize a grand opportunity to make computers say silly things. Dave Lawrence downloaded the list of sound effect names and trained a neural network to invent new names. Read the results here, here, and here. Some of my favorites that Dave generated:

Approach of piglets
Footsteps on ice, ponderous, parakeets
Fairgrounds: Ghost train ride with swords
Man punting metal hand
Waterfall into cattle bread

Unfortunately, we don’t know what these sound like, since it just generated the names of the effects. Now, it’s possible to train a neural net to generate entire new sounds, but I did something considerably simpler: I trained a text-only neural net to make up a new name, and then pick one of the 16,000 existing sounds to go with it. (link to dataset)

How well did it work? Well, the neural net did learn to choose valid existing sounds. I had to retrain it with a smaller, more interesting subset of the sound effects, because everything ended up being horses and heavy machinery. What you see below is a mix of results from both training runs. Click on the name of any of these, and it’ll play the sound the neural net thought should go with it. (Click on the number to find out the original name of the sound)

NOTE: sound will play as soon as you click the link.

07037122.wav Blinks.
07060061.wav 22 o'clock
07022197.wav German household operating.
07072020.wav Small small children
07005132.wav Piglet country.
07042121.wav Telephone, with chickens
07005121.wav Birds on mixer.
07042045.wav Hound up.
07026006.wav Interior, four o'clock.
07045119.wav Household 2 man barks.
07005205.wav Agitated door cat, interior, chickens - 1972
07037347.wav Cars: 1980.
07037445.wav 1 woman walking (reprocessed)

When the neural net’s sound effect is weirder, it’s harder to say whether it’s right or not. I’ve never heard any of these. So… maybe?

07065152.wav Birds of thunder.
07037379.wav Sheep operating.
07042196.wav Horse o'clock.
07050158.wav London down - traffic closed.
07037496.wav Small man continuous large poop.
07005206.wav Gravel bears - 1967
07064036.wav Flying people, 10 men, interior, applause - 1984
07039214.wav Telephone women, individual mews.
07066034.wav Electric water.
07038073.wav Many Men and some thrown.

07023107.wav Horses singing
07065046.wav Person stork.
07037366.wav National Parrot Road.
07071022.wav Death Interior, exterior, diosel notes (reprocessed)

And I guess we’ll just have to trust the neural net on these.

07005073.wav Firewomadoellic Bear departing.
07002266.wav Horse hopping on bonged screet.
07045229.wav Dinghy passes away.
07032091.wav Infant of ground.
07032270.wav Warble Yarring hour.
07010012.wav Scoop chimestimes bling. (Stolling Ghorters)
07058161.wav Electric School train seven o'clock (saying crush.)
07023217.wav Steel sparrows activity & two machine work, suburb passing over machinery.
07068025.wav Flying rubber sea.
07065052.wav Peacock butter, with background clock children.
07070107.wav Sixty Bubble Machine, 1967
07071043.wav Sawing brain dumping on bus, bombs women run.
07064008.wav Tempressed bow, rush of cows from machine with continuous singing
07039077.wav Lose Timber Machine of Button Transpoop opened

I also trained the neural net with the sound files and the names reversed - thus, I can finally ask it to pick a sound file to go with anything I want. Behold, long-standing mysteries solved by advanced artificial intelligence!

07042194.wav One hand clapping
07018034.wav Silence
07042215.wav T. Rex
07018033.wav The Beginning of the Universe
07005137.wav The music of the spheres

Support AI Weirdness and get bonus content: more sound effects (including the more PG-13 examples).