Re: [linux-audio-dev] applying RIAA curves in software

On Tue, Oct 25, 2005 at 12:24:39PM -0500, Jan Depner wrote:

> > If you download JAMin you'll find that I've already put the RIAA
> > curve in the sample .jam files. I've used this to record directly from
> > album to 24/96 with no preamp. It works fine.
>
> Wait, let me rephrase that - with no RIAA preamp. I do use a preamp
> for the signal.

I don't want to bring down the great work done by Jamin's authors, but
if fidelity is among you goals then using Jamin for RIAA equalisation
would be a bad idea.

First, Jamin's FFT based filtering can be quite inaccurate at low
frequencies. For example, push up the 31 Hz slider and measure
the result. This is because of the limited resolution of the FFT
at LF, and you can probably get around this problem by trial and
error.

Second, the general 1/F slope of the RIAA curve when combined with
cartridge equalisation should produce a -90 degrees phase shift
over most of the audio band. When you synthesize such a curve with
a multiband EQ, you will not have that phase shift, and any percussive
or transient sounds will be affected. You can easily hear this on
some instruments.

It's quite simple to make an RIAA filter with for example two
first order lowpass LADSPA plugins. Put them in parallel, set
one to 50 Hz and the second to 2120 Hz. Mix the outputs in the
ratio 9 to 1, and the result will be a *perfect* RIAA filter.

-- 
FA