I am new in this field (junior year) and I have a question may a little lazy for the majority of the forum.

I have a speech signal $ x(t)=0.4\cos(100πt)+0.3\cos(1800πt)+0.2\cos(980πt)$ and I know that the sampling frequency is $F_s=1000$

If I want to find the signal form after sampling the only thing I have to do is replace $t$ with $κT$

$ x(κT)=0.4\cos(100πκT)+0.3\cos(1800πκT)+0.2\cos(980πκT)$ but $T=1/1000$ so $x(k)= 0.4\cos(0.1πκ )+0.3\cos(0.18πκ )+0.2\cos(0.98πκ )$

Is the solution and the notation OK?

  • $\begingroup$ Seems normal to me $\endgroup$
    – Rizias
    Jan 28 '15 at 16:40
  • 1
    $\begingroup$ Have you heard of the sampling theorem? How do the frequencies of the sinusoids compare to the sampling frequency? Will there be aliasing? $\endgroup$
    – Matt L.
    Jan 28 '15 at 17:03

First there was a mistake in the simplification - when you divided by $T$. It should be $x(k)= 0.4\cos(0.1πk )+0.3\cos(1.8πk )+0.2\cos(0.98πk )$.

Notice that you sampling frequency is 1000 Hz, but your signal contains frequencies: of 50, 900 and 490 Hz. Therefore the 900 Hz will be aliased. To see what frequency it gets aliased to consider $$\cos (2\pi900nT) = \cos (2\pi900nT+mnT2\pi)$$

if $mnT$ is an integer, but $n$ is always an integer and $T$ is fixed by your problem. Therefore we require $m=\frac{p}{T}$, where $p$ is and integer. In other words $m$ must be a multiple of your sampling frequency.

In this case let $m=-1000$ then we have $$\cos (2\pi900nT-1000nT2\pi)= \cos (2\pi(900-1000)nT)$$ Performing the final simplification we get: $$ \cos (2\pi900nT)= \cos (-2\pi100nT) =\cos (2\pi100nT)$$ Note - the last equality just uses the properties of cosine - even function. Therefore your 900Hz frequency looks exactly like a 100 Hz tone when sampled at 1000 Hz.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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