Vladimir Vassilevsky wrote:
>
>
> Oli Charlesworth wrote:
>
>> This is a question which probably boils down to "what am I missing?".
>
> This boils down to N versus log(N) operations, exactly like with the
> FFTs. Look into the classic books such as "Multirate Signal Processing"
> by Rabiner.

I have "Multirate Signal Processing" by Harris, and I've thoroughly read
"Multirate Systems and Filter Banks" by Vaidyanathan in the past. Would
you recommend Rabiner's book as a useful addition?


>> It seems to be commonly stated that for interpolation or decimation,
>> it's generally more computationally efficient to perform the operation
>> in multiple stages, i.e. factorise the rate-change factor, and design a
>> lower-order filter for each stage.
>
> Athough this is generally true, this doesn't account for the overhead
> associated with the setup of the each filtering stage. The overhead may
> or may not be substantial.

Agreed. In particular, two separate filters require significantly less
storage for delay lines and filter coefficients than a combined approach.


--
Oli