I might have asked something similar long ago, but I don’t remember what the answer was, and perhaps things have changed for upcoming 7.0 so here I go again:

Would it be possible to add functionality for allowing a different sr in user opcodes?

The reason I ask is that I have a use case where I would like to be able to use some fairly steep, narrow band filters for low frequencies, and it’s difficult to get stable results without reducing sampling frequency. I was thinking that a first implementation could be a simple 2, 4, 8 etc factor reduction of sr.
… perhaps this could be done using an externally generated impulse response and a straight up multiplication using arrays and fft opcodes?

Regards,
Anders

Potentially it's already possible if you run the same filter twice or more times over the same input data using a loop and then taking the single last output of the filter.

That's a specific type of oversampling, double, triple, quadruple sampling. The filter parameters will need to be scaled accordingly because of the higher sr.

Also since Csound manipulates signals directly, any oversampling should be possible "by hand" too, implementing the upsampling and downsampling yourself,
and possibly the filter if needed. It won't be as efficient as in C, but it will work.

I thought Oeyvind had worked on this before, but maybe it was something else. I would need to study oversampling myself before I ventured an example beyond the simple case of double etc sampling

The whole area of UDOs needs a redesign to incorporate upsampling but it's possible and in our roadmap.

Prof. Victor Lazzarini
Maynooth University
Ireland

> On 9 Jan 2022, at 09:33, Anders Genell <anders.genell@gmail.com> wrote:
>
> *Warning*
>
> This email originated from outside of Maynooth University's Mail System. Do not reply, click links or open attachments unless you recognise the sender and know the content is safe.
>
> I might have asked something similar long ago, but I don’t remember what the answer was, and perhaps things have changed for upcoming 7.0 so here I go again:
>
> Would it be possible to add functionality for allowing a different sr in user opcodes?
>
> The reason I ask is that I have a use case where I would like to be able to use some fairly steep, narrow band filters for low frequencies, and it’s difficult to get stable results without reducing sampling frequency. I was thinking that a first implementation could be a simple 2, 4, 8 etc factor reduction of sr.
> … perhaps this could be done using an externally generated impulse response and a straight up multiplication using arrays and fft opcodes?
>
> Regards,
> Anders