- From: Raymond Toy <rtoy@google.com>
- Date: Wed, 11 Dec 2013 09:37:08 -0800
- To: Marcus Geelnard <mage@opera.com>
- Cc: Chris Wilson <cwilso@google.com>, "public-audio@w3.org" <public-audio@w3.org>
Received on Wednesday, 11 December 2013 17:37:36 UTC
On Wed, Dec 11, 2013 at 12:25 AM, Marcus Geelnard <mage@opera.com> wrote: > 2013-12-10 19:26, Chris Wilson skrev: > > Is there a historical reason why AnalyserNode has getByteTimeDomainData >> but no getFloatTimeDomainData? For higher-precision snippets, it would be >> kind of useful. For example, I've implemented an auto-correlating pitch >> detection in my pitch detection sample, and I think it would be higher >> precision if I could grab float samples (since I don't really need to >> update constantly, I wouldn't need to process ALL the samples through a >> ScriptProcessorNode). Ditto for the waveform display in >> http://webaudiodemos.appspot.com/oscilloscope/ - where I may want less >> than constant waveforms, but I want to grab higher precision than 8-bit >> data. >> > > I would also like to know why there's no float version (it shouldn't be > hard adding the function though). > > Also, is there a use case for the 8-bit data versions? Only possible idea > I've come up with is that you can use it directly as a WebGL texture source > or something, but I'd still rather see a high resolution version in the > first place (you can do the conversion to 8-bit yourself, but the other way > around is impossible). > > I did a quick peek through the blink history. It looks like it was born this way. +1 for a float version. > /Marcus > > > -- > Marcus Geelnard > Technical Lead, Mobile Infrastructure > Opera Software > > >
Received on Wednesday, 11 December 2013 17:37:36 UTC