On Thu, Mar 22, 2012 at 4:36 PM, Jussi Kalliokoski <
jussi.kalliokoski@gmail.com> wrote:
> On Thu, Mar 22, 2012 at 11:27 PM, Raymond Toy <rtoy@google.com> wrote:
>
>>
>
> Do you have a specific need for a non-power of two analysis? I'm just
>> curious.
>>
>
> It just seems like an arbitrary limit. Most of the time, you probably want
> to set the length in seconds, not samples, and then you won't get a
> power-of-two.
>
Yes, I'll agree that it's an arbitrary limit. And, yes, seconds (or more
lilkely, fractions of a second) is probably more useful and scales better
if the sample rate changes.
>
>
>> Also, FFTW uses the GPL license, so if the browser is also not GPL, there
>> are potential legal issues. This is where you go ask a lawyer.
>>
>
> There are other ones as well (kissfft is BSD 3-clause for example, IIRC),
> but I don't think the licensing of certain FFT implementations is a valid
> argument to base the spec on.
>
No, but you mentioned using FFTW in particular which has some licensing
issues.
>
> Also, the algorithm isn't complex, you could implement it without a
> library.
>
Agreed. But getting a fast one is complex.
Thanks for the info,
Ray