BiquadFilterNode seems underspecified

[rtoy]

The description of the various types of biquad filter nodes seem underspecified. Even for the ones that have a wikipedia link seem unclear and some (lowshelf, highshelf, peaking) have no references at all.

Perhaps we should just add the formulas from Rick Lyons (?) page from which these filters were derived.

[rtoy]

See http://www.w3.org/2014/03/06-audio-minutes.html#item06.

[padenot]

I did some work on that in the plane, you can look at it at http://padenot.github.io/web-audio-api/#filters-characteristics

[rtoy]

This looks really, really nice! The only thing I would add would be the formula for the transfer function. It makes the formulas clearer and pretty much does away with having to spell out the actual implementation algorithm since you already said the initial filter state is all zeroes.

[padenot]

On Sat, Oct 25, 2014, at 04:36 AM, rtoy wrote:

This looks really, really nice! The only thing I would add would be the
formula for the transfer function. It makes the formulas clearer and
pretty much does away with having to spell out the actual implementation
algorithm since you already said the initial filter state is all zeroes.

Makes sense, yes.

[rtoy]

I should have clarified it looks great in FF, but not so much in Chrome which doesn't have mathml. I thought we were going to mathjax for this.

[padenot]

This now looks good here on Chrome 38/Linux, sorry about that.

[rtoy]

Ah, beautiful. One last comment for now. I have not yet checked the formulas but the formulas use a plain alpha, but there is not definition for alpha. Presumably each function must be using of of the defined subscripted alpha variables.