The Science of the Voice

25 February 2008

Interview with 

Professor David Howard, University of York


David - Good Evening, Hi.

Chris - Thank you for coming down and joining us.  Your research, very interesting, is actually on the voice: something which we on the radio can't do without.

David - It certainly is and, of course, is something nobody can do without in this day and age.

Chris - So how do you go about studying it?

David - Well, there are a number of things you can do.  The most obvious thing is you can stick a microphone in front of people and look at the wiggles and try and make some sense of them.  That needs to be done in combination with understanding how we create a sound.  What are the key parts of our bodies we use when we create a sound?  One of the things we've been looking at in the past is differences between people who sing or speak.  For example, we've been looking at choristers in cathedrals to try and establish whether people can tell the difference between them singing and also to try and establish what it is that changes as boys and girls grow older.

Windsbach Boys ChoirChris - So you've given me some samples.  Let me invite the audience to participate in a bit of a challenge then.  You've given me two groups of choristers: one all boys, one all girls singing the same piece of music and we're going to see if we can get people to tell the difference.

David - This is correct.  The key thing when you're listening is to realize this is a choir singing so you need to listen to the top line of the choir.

Chris - So I'll pick one of these at random so here we go.

Choir Sample 1

Chris - Any thoughts Helen?

Helen - Just enjoying it, it's rather lovely.  I don't know, I think I'm going to have to hear the next one to compare it to.  It sounded fairly female to me.

Chris - Ok, should I play number two David?

David - Yeah, try number two.

Choir Sample 2

Chris - I have to say I cannot hear the difference myself and I've done quite a bit of singing in a lot of cathedrals and choirs.

David - Well I have to confess that I too have and listened to many children whilst making these recordings.  I find it difficult to tell the difference.  If you ask me to tell the difference I would struggle.

Chris - So if I let the cat out of the bag, the one I played first was the boys singing and Helen thought it was girls.  The second was the girls singing.  Does this then put the kibosh on this claim that you have to have all male choirs and especially all male treble choirs (boys) because they sound better than girls?

David - Well I believe it does.  What we're demonstrating here is that given the same situation boys and girls can fulfill the role and do a job as good as each other.  But it does come from other areas too because boys have been singing in cathedrals now for over half a millennium and girls have been singing since 1991.  So they haven't got much of a head start in terms of catching up.  This does show that at Wells Cathedral, which is where these recordings were made.  Actually girls can do the job as well as boys.

Chris - And if you feed those recordings into a computer and ask it to build you a profile of what the voices actually look like does that give you any obvious standout differences between the boys and the girls?

David - No it doesn't.  Interestingly, our ability to analyze things for the real subtlety of listening we cannot yet with a computer do as well as our ears.

Chris - Which makes a change doesn't it?

David - It does and it's rather nice I think that we haven't yet beaten the ear.

Chris - Ok, so let's zoom in on the business end: what's doing this?  The whole vocal tract?  Can you just give us a whistle-stop tour of this and how it works?

LarynxDavid - Well very briefly, if you're going to make any sound you need a source of sound.  For most of what we do when we're speaking it's the vocal folds which vibrate inside the larynx.  So that lumpy thing on your neck: the Adams Apple at the front that you can feel with your finger.  If you put your finger on it and swallow it moves up and down to show that it's alive and well.  There are two muscles in there that vibrate and they produce a buzz.  Now I've got an electronic version of a buzz here so this buzz you're about to hear is an electronic larynx.

Chris - Not a shaver?

David - No, it might sound like one. Here we go.

A voice-box simulating buzz

So that's a buzzing sound and to prove that this simulates what your or my larynx are doing, I'm now going to tell you my name and use this instead of my normal larynx by placing it against my Adams Apple.

Hello my name is David Howard - as spoken using a larynx simulating buzz

Chris - So how's that working?  You're just sending a pulse of sound waves in through your throat and it's hitting your vocal chords.  You're not breathing and talking as you would do normally, you're just moving your voice box as though you were going to?

David - That's correct.  The way you use this is to hold your breath.  So you take a breath and go, 'ha,' as if you're going to lift a heavy weight which closes the vocal folds and jams them together so that you can build up pressure in the lungs to support yourself.  It also means when you put the buzz in it then stimulates the cavities above in the normal way.  The key to speaking is that we have two things available to us.  We've got a buzz and we've got a tube and this tube is a squidgy tube which goes from the larynx to the lips, the mouth.  And of course, when we're talking we're moving our jaw, our tongue, our lips and that changes the shape of the tube and therefore changes the acoustic characteristics of that tube.

Chris - So can you model that.  If you take measurements from somebody and record over time can you build up a profile of what their throat looks like so it's possible to recreate their voice almost?

David - You can and this is something we're now doing.  This is looking at how can we electronically synthesize a natural sounding human, starting with the real tube shape of a human being.  You put them into a Magnetic Resonance Imaging Machine and if you do that and characterize the tube: I have in front of me two acrylic tubes.  This one is shaped like the vowel, 'aah.'  It's 17 and-a-half centimeters long and it sounds like this. I've put the buzz in the end.

The sound created by a larynx simulator and acrylic tube modelled on the throat whilst making an 'R' sound.

Chris - So how does that work?  How are you producing that particular sound?

David - Well, the way the sound is produced is because the tube itself has an internal shape which is modelled on a real human mouth saying, 'aah.'  So if you imagine saying 'aah,' you open the jaw.   If you contrast 'aah' with 'ee,' 'ee' has a closed jaw.  'Aah' has an open jaw and I also have a tube for 'ee' which sounds like this:

The sound created by a larynx simulator and acrylic tube modelled on the throat whilst making an 'E' sound.

Chris - Sounds like it went down a bit at the end there, maybe it's getting tired or something?

David - The pitch goes down because the buzz goes down.  The key thing is that the shapes are producing 'aah' and 'ee' and what I like to remind singers, if we're talking to singers, is that all you have to sing with is a buzz and a tube that's about 17 and a half centimetres long which is all squidgy, that's it.

Chris -   Well, let's look at something which people who make their money from the voice: people who take the rip out of other people and sound like people.  How are they doing that because that all comes down to the same business as to why we have accents?  I know we're going to have a flurry of 5 million emails as people love the subject of accents.  Why do I sound English?  Helen sounds English yet a lot of people on the other side of the Atlantic have their own distinct way of sounding.

David - We have the basic speech production apparatus.  We all have the same larynx and the same tubes above it but culturally depending on where we're brought up, what are the sounds we hear when we're very young?  That begins to determine what we're going to sound like when we speak, and indeed sing, and indeed make music.  So if you're trying to imitate, if you're a professional imitator: like, for example, Rory Bremner.  He will take a voice and we will try and imitate that voice but he's got a fundamental problem.  The fundamental problem is that he's got to do it with his tube and apparatus and he will not have the same tube as the person he is trying to imitate.

Tony BlairChris - Now you spent some time with Rory Bremner at the BBC, he's often on telly doing this.  He was taking the rip out of Tony Blair.  So you've given me a lump of Tony Blair.  Here's our good old friend Tony:

"It's not the consequence of foreign policy, it's an attack on our way of life" - Tony Blair

Chris - Everyone remembers that speech very well.  Here's Rory Bremner doing Tony Blair:

"It's not the consequence of foreign policy, it's an attack on our way of life" - Rory Bremner as Tony Blair

Chris - Now what's really interesting is that I can tell that's not Tony Blair but at the same time it does sound dramatically like him.

David - It does.  The acoustics are different in both cases.  Tony's was in a big hall and it sounds like a speech.  Rory was in a quiet studio.  He will never produce exactly the same acoustic signature.  He can't because of the shape of his own tubes are different to Tony Blair's.

Chris - So why is he fooling me to a certain extent then?  How do I know who that is instantly?

David - Well the key thing he seems to be doing is latching onto some key acoustic characteristics that change when Tony Blair speaks in the way that Tony Blair does it.  They're kind of his acoustic mannerisms that he uses when he speaks.  What Rory is doing is picking those up, rather like your caricature artist might draw a cartoon with a big nose, and you know who it is even though the actual picture is a circle: two dots and a big nose.

Chris - Well thank you very much, David Howard from the University of York.


Add a comment