Re: fftSize requirement for RealtimeAnalyserNode

Thanks for creating the issue Al!

Not just performance, but yes! :)

Jussi

On Thu, Mar 22, 2012 at 5:28 PM, Alistair MacDonald <al@signedon.com> wrote:

> Issue-7 created:  https://www.w3.org/2011/audio/track/issues/7
>
> Should arbitrary FFTs sizes be possible in the RealtimeAnalyserNode?
> Currently they are limited to a Power of Two. Jussi suggests we will get
> better performance from the API if we changed this.
>
>
>
>
> On Thu, Mar 22, 2012 at 7:07 AM, Jussi Kalliokoski <
> jussi.kalliokoski@gmail.com> wrote:
>
>> I hadn't actually even noticed this. Interesting. Why exactly should the
>> value be a power of two? It seems to me like the spec shouldn't say
>> something like this. Sure, power of two FFTs are easier to implement than
>> others, but IIRC they are rarely the fastest. And if we aim for
>> performance, the implementation should probably be using something like
>> FFTW that allows for any buffer size anyway.
>>
>> Cheers,
>> Jussi
>>
>>
>> On Thu, Mar 22, 2012 at 11:02 AM, Wei, James <james.wei@intel.com> wrote:
>>
>>>  ** **
>>>
>>> For fftSize in RealtimeAnalyserNode in the spec, it only says the value
>>> must be power of 2.****
>>>
>>> ** **
>>>
>>> But in fact in the implementation, the value must be limited in the
>>> range of [128, 2048], and it will throw the exception if illegal value of
>>> fftSize is set.****
>>>
>>> ** **
>>>
>>> Which should be documented in the spec. ** **
>>>
>>> ** **
>>>
>>> ** **
>>>
>>> Best Regards ****
>>>
>>> ** **
>>>
>>> James ****
>>>
>>> ** **
>>>
>>> ** **
>>>
>>
>>
>
>
> --
> Alistair MacDonald
> SignedOn, Inc - W3C Audio WG
> Boston, MA, (707) 701-3730
> al@signedon.com - http://signedon.com
>
>