Re: Suggestion: Web Audio graphs that does not output sound but can write to an AudioBuffer instead

On Mon, Apr 16, 2012 at 12:10 PM, Raymond Toy <rtoy@google.com> wrote:

>
>
> On Fri, Apr 13, 2012 at 1:47 AM, Per Nyblom <perny843@hotmail.com> wrote:
>
>>  Hello,
>>
>> it would be great to be able to access the generated sound data without
>> having to send it to the speakers as well.
>> This feature is very useful for a musical application that should render
>> a wav-file or similar (perhaps with the help of the FileSystem API). Most
>> existing musical applications support this "offline" rendering mode.
>>
>> It is also useful to be able to create AudioBuffers as an optimization of
>> a graph. Suppose that you create a very large graph that generates a sound
>> effect or musical instrument sound and want to reuse it. It is very
>> convenient to be able to generate AudioBuffers to improve performance
>> without having to do this with an external program.
>>
>> All this could perhaps be supported by using a subclass of AudioContext
>> that supports methods like: renderToBuffer(AudioBuffer, bufferOffset,
>> length) or something similar.
>> It is important to be able to incrementally render to the buffer because
>> of the single-threaded nature of JS (you can use Workers for this but I
>> think it is important anyway).
>>
>
> Won't a JavaScriptNode work for this, where the node just saves the data
> away in an audiobuffer?  Or are you saying it won't work because JS is
> single-threaded?
>
> Ray
>

I think the idea of an OfflineAudioContext is what we want, because ideally
the rendering will occur faster than real-time.

Chris