1
$\begingroup$

I am trying to understand and find a way to distinguish two same sounds of different people by some physics formula, so could you guys help me?

OK I'll try to explain my question in this way that, for example, there are two people A and B are reading this sentence aloud. "Crackers are Cracking for Crown". And if we want to find difference by applying any kind of physics, maths or computer science formula, so and then later we can say this was person A sound or B etc.

But while finding the difference these points should not matter:

1- volume of the sound
2- pronunciation (obviously it understandable)
3- speed of the sound 
4- same voice or different
$\endgroup$
  • 2
    $\begingroup$ Have you ever heard of voice-recognition algorithms ? $\endgroup$ – kaka May 21 '14 at 3:26
  • $\begingroup$ You will have more luck in SP.SE. $\endgroup$ – Davidmh May 21 '14 at 6:19
  • $\begingroup$ @Kaka: Yes I have heard, but I'm researching physics so I can implement in my computer program. $\endgroup$ – c-sharp May 21 '14 at 13:44
  • $\begingroup$ Related: physics.stackexchange.com/q/78827/2451 and links therein. $\endgroup$ – Qmechanic Jun 6 '16 at 7:19
  • 2
    $\begingroup$ I'm voting to close this question as off-topic because it is not about physics. $\endgroup$ – Emilio Pisanty Aug 9 '16 at 13:18
1
$\begingroup$

There are many algorithms used to analyze audio waves and voice sound in particular. I don't think you will find a single "Physics Formula" that can quantify or distinguish between two similar audio samples. Algorithms exist for what you are describing, but I'm not sure if any good ones are open-source or published.

Voice recognition algorithms attempt to match audio samples with words, regardless of the pitch and speed of the voice. A voice biometric algorithm attempts to match a particular speaker (person) to an audio sample.

Both of these things are very complicated and are areas of active research. We all know from experience with Smart Phones and phone prompt systems that voice recognition is not very accurate. But with a limited vocabulary, the algorithms do a good job (for instance if we know the user will either say "yes" or "no" the system can distinguish between these two words accurately) I don't know how accurately computers can match a voice to a person, but since it is used as a security measure, It should be pretty good. (see this biometric update) Google is actively researching ways of improving speech recognition technology.

$\endgroup$
  • $\begingroup$ Thanks for the reply. I was looking for a specific search term like you used "voice biometric algorithm". Actually I'm researching on Google talking robot and IPhone Siri algorithm. $\endgroup$ – c-sharp May 21 '14 at 13:40

Not the answer you're looking for? Browse other questions tagged or ask your own question.