What makes a voice 'yours'?

I was playing around with sound recorder (Programs, accessories, entertainment, sound recorder on Windows) and discovered that you can also play what you have recorded in reverse (a dropdown from 'effects'). Two things struck me, How surprisingly unintelligible my speech was when played backwards (even a basic phrase like 'good morning'). Secondly, although unintelligble, it was unmistakably me. What is it about the human voice, with its very modest number of phonemes, that makes it virtually (totally?) unique - and recognisable, even when played backwards?

Overtones.

Every person's and every musical instrument's sound is composed of waves of different frequencies, whose combination and interplay give the sound its color and recognizability.

I don't know if this is still the case what with modern phones and all, but I learned in high-school physics class that while a person's voice can be composed of dozens if not hundreds of different frequencies, and recognizing a voice as the voice of someone you know may require about five or six of the most prominent ones, telephones transmit only the most conspicuous wavelengths and filter out all the rest. As a result it is usually possible to recognize someone you know when speaking on the phone, but not always, and distinguishing between e.g. three sisters who speak with similar voice quality may not be possible, because it requires more information than the transmitted wavelengths provide.

And what makes the overtones different, you ask?

Size, structure (bone density for example), and shape of the vocal cords, larynx, mouth, throat, sinuses, head, chest, etc.

Pretty much the same kind of thing that makes a guitar sound different from a lute or a trumpet from a cornet.

Pretty much the same kind of thing that makes a guitar sound different from a lute or a trumpet from a cornet.

not to mention, what makes a Stradivarius sound different from a "john doe," or even one Strad from another.

In the book “Blink” Malcolm Gladwell writes about how psychologists conducted experiments to find out why certain doctors were more likely to be sued. They recorded doctors talking with patients and removed the high frequency sounds that enable individual words to be recognised. From the remaining garble intonation, pitch and rhythm are preserved as well as human qualities such as warmth, hostility etc.

From listening to these recordings they were able to predict which doctors were more likely to be sued.

Here’s the extract:

http://www.theleadershiphub.com/files/Whydoctorsgetsued.doc

That’s not exactly what you’re talking about but it might interest you.

I’ve spent hours and hours cutting up voice samples on computer, mostly singing, and it is quite an eye-opener.

For one thing, if you cut up an individual word it’s often totally unrecognisable out of the context of the sentence.

As a made up example, if you take the phrase “I never felt like this before” and cut out the word “never” you might find it actually sounds like “ner”. It’s impossible to get a recognisable, separate “never”. Then listen to the whole phrase again and there it is “I NEVER felt like this before” as clear as day.

From this I conclude that it’s your knowledge, prejudice, of listening to English for years that “fills in the gaps” and enables you to hear words that aren't actually properly intonated. Recognising unclear lyrics from songs sung in languages you are not absolutely familiar with must be very, very difficult.

Another phenomenon is if you try and create a sample that begins with, let’s say, the word “stop”. It sounds like “top” so where’s the “s”. You then find a tiny, tiny squiggle on your screen in front of the “top” that is a barely audible “sss”. You need that almost inaudible “sss” to make the word, without it there is no “stop”. From this I conclude that our brains are incredibly sensitive to small changes in the sound of a human voice. Not exactly a revolutionary idea, but it’s interesting to see the effect for yourself.

#5 -- The t in top is quite different from the t in stop. (The t in top is much more strongly aspirated.). So it may be that the main factor that enables English speakers to distinguish stop and top isn't the tiny squiggle of an s but the very different character of the t. It's not a subtle difference, it's just one we don't think about.

Edited by: VinnyD

4 -i can't read all that. so, even tho i am not a doctor -what is the way to talk not to get sued? is there anything i can do about it, or are some people just made the suable way?

That's really interesting about the doctors, Count Zero. Thanks for posting the excerpt.

Mathilda, it's not a long article. But in short: people don't want to sue doctors they like, and it turns out that that unlikableness can be predicted from short speech samples from which the content has been removed. Mainly a question of sounding 'dominant' vs. sounding concerned about the patient.

Re#5 - way back in the 1960s I did some work with voiceprints. We made recordings in an anechoic chamber (really creepy) and were analyzing for all sorts of things, including word-boundaries, and the contextual influence of various phonemes.

We spent a lot of time on consecutive vowels, and extra glides that creep in between words.

e.g. "very active" - there isn't a "y" (or "j") glide in either word, but it creeps in here, unless you are a singer or a radio announcer and insert a slight pause.

We used a lot of Navy research on communications channels and how much redundancy you really needed for clear understanding. You definitely needed some, but not as much as you might think. Environments of limited vocabulary and standard situations made matters much easier, of course.