help with signal processing

I have 5 different signals, and several subjects. I'm acquiring data of these 5 signals for each one of them, and I need to extract feactures from these signals.

I'm getting buffers of 256 samples with 64 samples overlaping for each signal. For getting an accurate result, I should extract the features for each buffer, or can I just extract features of each signal even though they have different lengths?

[Image Analyst]

Gee Catalina, thanks a lot for wasting my time.

[User 864]

I was just making the question clearer.

Answer 1

[Image Analyst]

" but for the feature extraction I need the features of the whole signal" -- What does that mean? You have the whole signal, and you pass it into your function featuresFromBuffer(), and you get back a feature vector (presumably of a fixed length). In the loop, you're eventually having the function look at the whole signal, a chunk ("buffer") at a time. So what's the problem?

[User 864]

I mean, I have signals of different lengths (some last 40 seconds, others 30 minutes, etc.. ), for getting an accurate result, should I extract features for each buffer, or for each signal no matter their length?