the IR is not resampled when the SR != 44100

[chikashimiyama]

the IR data is sampled in 44.1 k according to
https://sound.media.mit.edu/resources/KEMAR.html

Thus, this data doesn't produce proper results when the data is used in another sample rate.
The only ways to solve this problem is to

1. prepare another set of ir data
2. upsample / downsample the existing data

@danomatika

[danomatika]

@Spacechild1 Do you think up/down sampling would be difficult to implement?

[Spacechild1]

I have no idea :-)

[umlaeute]

I think many frameworks that work with IR sets will refuse to do resampling. e.g. SoundScapeRenderer, and probably for a good reason.

So I do not think that it is feasible to bring this to earplug.

probably earplug should refuse to use the built-in IR set if it detects the wrong samplerate though, and require the user to to provide their own.
i find it unfortunate that the IRs are provided in a custom(?) textfile, rather than some more standard format (e.g. a multichannel soundfile, or a SOFA file), which would allow the object to check whether the samplerate is correct.

[danomatika]

We could reformat the IR set.

[umlaeute]

We could reformat the IR set.

what do you mean?

[umlaeute]

here's the issue of SoundScapeRenderer I had in mind: SoundScapeRenderer/ssr#32

[danomatika]

I mean reformat the existing data values into an audio file or other format as you wrote earlier, then adjust how the file is loaded. It would then be compatible to load other sets but still work as it is by default.

[umlaeute]

thanks for clarifying