(off topic?)RIFF/WAV by Frame/sample and Multople Frequencies in same Frame/sample

[lzf]

So I poked around a bit but couldn't find any place this looked at home, since some of it (on multiple frequencies at the same time) I figured this might be the best bet. I apologize if it is too far off topic, please let me know what would be a better home for it and future questions of the same type.

One of the audio projects I'm working on is going to need to deal with RIFF/WAV data (PCM 16byte, mono, ≈44k sample rate). Most of what I've found so far has been on creating live wave forms, rather than writing to files and I'm looking for some specific issues to these.

What I've got is a 2-dimensional matrix of frequencies. I want to convert this to RIFF/WAV data frame by frame since the frequencies are arbitrary and usually not related to that surrounding it. So what's the best way to 'paint' a WAV data sample/frame by sample/frame and not lose the sound I'm working with?

Secondly, some of these frames will thus contain multiple frequencies (so say A440 and 2400 both at the same time for the duration of the frame): how is this calculated? Are there any essays on this? The closest I found was something saying to take all the frequencies that were present and use the average of them all (so for (440 + 2400)/2) but this to me doesn't seem like it would produce a multiple tone, just a solid tone at 1420. Maybe I'm lacking some knowledge on acoustic math here but that would be a single distinct sound rather than two concurrent. So what should I even look up to try and find more info on how to do this? I've looked up polyphony but that usually results in other questions by other people that haven't been answered.

Thanks everyone! And I look forward to being a contributing (as well as quereying) member

[Nystul]

I can't speak on the technical aspects, but in order to establish a pitch there has to be some sort of periodic cycle established. In order for something to be 800 hz, it has to have some type of waveform that repeats 800 times per second. How long would it take to establish that? I'm not sure what the most correct answer to that is, but it is going to require relationship over time. It cannot be done with a single sample.

[DSmolken]

Are you describing spectral synthesis where you have a 2D graphic image of a sound which is then used to generate soundwaves?

[J.F.K.]

Use Sony's Spectralayers software. Or Photosounder. Photosounder has its forum here at KVR.

[lzf]

Are you describing spectral synthesis where you have a 2D graphic image of a sound which is then used to generate soundwaves?

DSmolken - that sounds similar to what I'm trying to accomplish. Will be googling and searching here, do you have any resources you'd recommend though?
Thanks!

[DSmolken]

If that's what you're after, then see above - Spectralayers or Photosounder should work. I use Harmor which also does image-based synthesis, but it does a million other things too.