Wikipedia Is Building a Repository of Famous Voices

Even if you don't get invited to the right parties, you can now hear a famous person introduce themselves to you in person thanks to a new project by Wikipedia. The site is encouraging anyone remarkable enough to have their own page on the online encyclopedia to submit an audio recording of themselves, in an effort to preserve the voices of the great, the good, the notorious, and the mildly interesting among us.

You can now hear the first “Wikipedia voice intros” on the relevant biographical pages of some 54 notable individuals, mainly in English, but with a smattering of French, Hebrew, Japanese, and Swedish.

British entertainer Stephen Fry was among the first to record his own personal introduction. At the bottom of the right-hand column, beneath Fry’s signature (presumably not the one he uses on his credit card), there’s an audio file in which he says, “Hello, my name is Stephen Fry, I was born in London and I’ve been in the entertainment business since, um, well, I suppose about 1981.”

All of the intros follow the same structure, and the Wikipedia page for the project explains the point is for posterity and pronunciation: “We do this so that we know what notable people sound like; and how they pronounce their own names.”

The project was put together by Wikipedia editor Andy Mabbett, and anyone who’d the subject of an article is invited to submit an audio recording like the above. “You can help the Wikipedia voice intro project by asking people you know who are the subject of Wikipedia articles to make recordings of their voices in any language in which they’re comfortable (the project is not just for English speakers),” explains a post on the Wikimedia blog. “You can also help to transcribe the existing files into timed text captions.”

For now the selection is pretty eclectic, featuring mainly British people from a wide range of fields. Most of them aren't exactly famous, but are notable for their work in various areas of expertise. You can listen to the careful speech of Essex-born botanist Alastair Culham, the bright tones of London Assembly member Jenny Jones, or the upbeat voice of Chair of Arts Council England Sir Peter Bazalgette (whose name I learned from this is actually pronounced “basil-jet”), for instance.

Quality varies as in most cases the subjects are recording themselves, but there’s something rather intimate about hearing people in the background or muffled sounds when someone’s too close to the microphone.

But perhaps even more exciting than the intro project is Mabbett’s adjacent effort to extract other people’s voices from BBC radio shows. He’s working with the broadcasting corporation to get snippets for the Wikimedia Commons library, which include notable people talking on different subjects.

At an event called the Speakerthon earlier in the month, the Wikimedia community and the BBC teamed up with the Open Knowledge Foundation and Creative Commons UK to capture 300 voices from the Radio 4 archive. At the time of writing, 133 had been processed and were available.

The result is again rather broad-ranging and often unexpected; I’ve picked out some of my personal favourites below, all via Wikimedia and the BBC through a creative commons license.

Aung San Suu Kyi speaks about nearly getting shot in an episode of Radio 4’s popular show Desert Island Discs:

Benedict Cumberbatch talks about sliding into a bramble bush in what turns out to be a recollection of riding a horse as a child:

Damien Hirst wasn’t as good at drawing dinosaurs as another boy in his class:

Margaret Atwood says poetry was always easier because you could “crank it out in your cellar”:

Morgan Freeman does not embody gravitas:

And there are plenty more where they came from. None of the clips are new, but a repository like this serves as an interesting reminder of times past. We’re so used to collecting photos and even videos from events current and past that audio sometimes becomes a forgotten medium.

According to the Telegraph, the BBC is also looking into using the resulting bank of information on Wikipedia as a “fingerprint” for famous voices—a dataset which could work with a voice recognition system to automatically tag archived clips. They’re releasing 500 to 1000 clips to get the collection started.