On 2011-01-17, Paul Hodges wrote:
With all this discussion of the accuracy of IIR-based encoders, where does the use of convolution using (for instance) Angelo's impulses stand - with respect to accuracy, and I guess also speed (it's fast enough for me, being about 8x realtime at 44.1kHz on the last processor I measured it on)?
Convolution is convolution. It's always rock steady, and perfectly reliable if implemented half-right. It's never as fast as a fully optimized, dedicated filterbank, because there's a whole ton of optimizations and approximations you can do with the latter that you can't with fully general convolution algorithms. But why would we care when it goes fast enough still and in the meanwhile buys us full generality?
Of course, there are lots of nits to be picked with fast convolution algorithms. The numerical accuracy kind in fully steady state LTI operation, and if modulated in any way, also (short term) stability issues (longer term ones do not surface with FIR topologies like these, by definition). Yet by and far, you can deal with those using extended arithmetic, or just rely on proven and numerically optimized libraries which take care of the problem for you.
-- Sampo Syreeni, aka decoy - de...@iki.fi, http://decoy.iki.fi/front +358-50-5756111, 025E D175 ABE5 027C 9494 EEB0 E090 8BA9 0509 85C2 _______________________________________________ Sursound mailing list Sursound@music.vt.edu https://mail.music.vt.edu/mailman/listinfo/sursound
