- From: Kevin Gadd <kevin.gadd@gmail.com>
- Date: Mon, 17 Jun 2013 09:51:54 -0700
- To: Russell McClellan <russell@motu.com>
- Cc: Marcus Geelnard <mage@opera.com>, "public-audio@w3.org" <public-audio@w3.org>
Received on Monday, 17 June 2013 16:53:01 UTC
If you're going to expose a reusable FFT, there's an argument to be made that you might as well make it a commodity function so that it can be used on any stream of data in a TypedArray instead of specifically things that are being fed to Web Audio. After all, audio isn't the only place people use FFTs. -kg On Mon, Jun 17, 2013 at 8:56 AM, Russell McClellan <russell@motu.com> wrote: > > On Jun 17, 2013, at 11:29 AM, Marcus Geelnard <mage@opera.com> wrote: > > My original proposal was to rip the FFT part out of the analyzer node > (it can just as well give the time domain signal only), and expose the FFT > functionality in a JS interface instead. That way we would have the same > functionality and performance for the analyzer node (when combined with the > JS FFT interface), but the FFT primitive could be used in the > ScriptProcessorNode too for a wide range of interesting effects (not to > mention for using it for non-Web Audio stuff as well). > > > As I said just last week, I think even better would be a > "SpectralProcessorNode", which would act exactly like the > ScriptProcessorNode except on spectral data. You could control parameters > like overlap, window, and fft size, and these would have reasonable > defaults. I think just exposing an FFT function would be a little too > "low-level" in character compared to the rest of the web audio API. > > Thanks, > -Russell >
Received on Monday, 17 June 2013 16:53:01 UTC