I'm brand new at digital audio; please bear with me while I learn. I have a technical background so hopefully I'll prove to be a fast listener. Anyway, I'll do my best.

I've got foobar installed and functioning; I'm competent with its basic functions but pretty ignorant with anything more sophisticated. I would like to open a digital file (i.e., a wav file), apply a DSP effect (like a certain impulse response function in Convolver) and save the output to a new file that would be otherwise identical. If I apply an impulse function, for example, I would expect the copy to be identical to the original file. The idea is to prepare digital files for ABX comparisons.

Can I do this with foobar? If so, I'd be very grateful for a link/short tutorial. If not, what tool would I need to do this job?

Thanks,
Jim

Set up your DSP chain as you need it to be, make sure you have the diskwriter installed and check "use DSP" in its options. Pick or create a preset you want to use, then select your files in the playlist and right click > Convert > Run conversion.

You can use the diskwriter to a lossless format (e.g. wav or flac, so the only difference would come from the dsp applied, not from transcoding) and you can enable dsp-s in the diskwriter. If this is not enough, ask I'm just lazy to look up how it's done exactly.

edit: okay, I was late.

Ruby,


Done. And successfully I think. Thanks very much. One question, though. When convolver is active but no impulse file is loaded, or when unitpulse2k.wav is loaded, the output file is slightly different than the imput file. Filesize increases by 6kB. A comparison in EAC yields:

error type: 2048 repeated samples Position: 0:00:00.287 .

Output file is .032 seconds longer.

Now realistically I KNOW this effect won't be audible. But I need to make sure I can make that case. We're talking about a .032 second timing difference essentially right at the beginning (less than a second into) the file. Otherwise the two files are identical. Am I reading this right? (A related question: what's a good program to use to compare these two files visually?)

BTW: other impulse functions generate larger differences. That's not surprising.
[Edit: this isn't true. The functions that I loaded that generated larger differences weren't proper impulse functions. Other impulses yield different differences, but the change in the filesize is about the same as it is with unitpulse2k.] But I would expect the unit (true) impulse to result in an identical file.

Thanks again...

Jim

The delay (-> increase in length) of unitpulse2k is caused by its length. Create a 1 sample long unitpulse to create bit-identical output.

Thanks, devo.

Jim

This won't work because IIRC convolver has a minimum filter length that is bigger than 1 sample.

Thanks, Garf. Using Audacity I made what I thought was a proper (short) impulse--a single sample of unit volume--and ran my data through it and got an identical file back. I felt pretty good about that. Are you saying that it doesn't do anything under these circumstances, that the impulse file is just ignored?

Doesn't matter so much now; since I got Audacity installed I was able to compare the input and output files directly. The only noticeable effect of convolving with impulse2k was a very short delay (.045s) at the beginning of the file. Then again, I also realized that my test file--an older piano recording, has very little data above about 10 kHz. So I'm not sure 'bout that result anymore. Have to try it again with a more modern recording.

Question: is this--a short delay and nothing else--what you would expect when a file is convolved with impulse2K?

Thanks,
Jim

When an impulse is smaller than the minimum length, it will extend it internally with zeroes.

Yes, a small delay (and some extra silence at the end) is the expected result.

Theorethically roundoff errors could also cause differences, but the convolver has enough precision that they should not matter in practise.