Can we filter an int16 signal using filters FIR?

The coefficients resulted in the design of the filter FIR are float, and when I try to filter the signal it just can't. I tried with "conv", is there a way?

Best regards, Marcelo

[Jan]

What does "it just can't" explicitly mean? Pleas post the error message and define the input arguments exactly. The less we have to guess, the easier is an answer.

Answer 1

[Wayne King]

Jan is correct about making your post clear.

If you are using conv(), the help for conv() states that the only supported data types are:

float: double, single

So why not cast your input to double()?

conv(double(x),B)

where B are the filter coefficients.

Or better yet, use filter()

y = filter(B,1,double(x));

Answer 2

[Jan]

What about converting the signal to a double temporarily?

y = int16(filter(B, A, double(YourINT16Signal)));

Another idea would to use an M-version of FILTER, see e.g. Answers: Hard coded filter parameters

A C-Mex version of FILTER can be found in the file exchange and it can be adjusted to integer types. But the conversion to double seems to be the most elegant way, if it is not impeded by the memory consumption.

Answer 3

[Sangroku]

Sorry for not explaining well, and thank you for your help. And how do I get int16 samples from a file .wav? I tried:

>> [s1,fs] = wavread('C:\Users\user\Desktop\ex.wav');

>> s=int16(s1);

>> wavplay(s, fs)

t just noise this way

[Wayne King]

How is this at all related to your original post about filtering?

Have you tried reading the file with the 'native' option.

 [s1,fs] = wavread('C:\Users\user\Desktop\ex.wav','native');

What datatype does that give s1?