Signal Processing: differences between Compose and HyperGraph

Hello, Community!

I wanted to share an experience acquired while comparing Signal Processing functions of HyperGraph and Compose. Every now and then someone asks “why is Compose giving a different output in comparison to HyperGraph?”, but as a matter of fact most of the time they are comparing bananas to apples.

There are usually many differences between both functions when the default values are used:

• Sampling frequency
• Using a time-domain filter in Compose (such as filtfilt function) and using a frequency-domain one in HyperGraph (such as batch function)
• No zero-padding on the data in Compose, which is the opposite of how HyperGraph works

Although some users are convinced that HyperGraph is right, due to the long time the filters have been there, it is not a matter of right or wrong, but rather of different procedures.

When all differences are mapped and explained, such as in the bullet points described above, we reproduce batch function from HyperGraph in Compose:

HyperGraph's batch filter can be reproduced with the attached OML script, considering a bandpass filter in frequency domain.

In a nutshell, the main difference is usually that one filter is operating in time domain, whereas the other one is doing the same but in frequency domain.

Regards,

Roberta