"hacking" SAM to do convolution

[David A]

SAM basically covers the frequency range from 200 Hz on down. That's the bottom 3 and a half octaves of the 20 - 20,000 Hz range we regard as audible, roughly one third of the range. I don't know how many adjustments within that range the hardware that implements SAM is capable of handling but if you want to do room correction you want to cover the whole 20 - 20000 Hz range. That's going to require a lot more adjustments to do well than covering the range from 200 Hz on down.

The processor which applies SAM corrections may be capable of doing convolution processing but does it have enough power to do the amount of processing required to make an improvement over 10 octaves rather than the 3.5 octaves required for SAM? You probably aren't going to get by with any fewer adjustments from 200 Hz on down in a full range room correction filter than you do with a SAM profile so if you hack the SAM processing code you're going to ask it to handle the corrections for the current 3.5 octave range it is intended to handle plus the corrections for another 6.5 octaves.

I'd want to know a lot more about how SAM is implemented, how much of the processing power in a Devialet is devoted to that processing, and whether it's a separate processor or just a routine in the overall processing code.

Like everyone here I think being able to implement room correction in the Devialet would be of great benefit but I don't think hacking the code for processing SAM sounds like a good way to go about it. I think room correction is going to take more processing muscle than is used for processing SAM profiles and if you can hack the SAM code and access enough processing power to do what you want you may well be taking processing power away from other essential functions which might actually make your results worse rather than better.