Syntrillium Backup Forums :: View topic - Paste->Mix with logical AND

swingheim · Posted - Fri Mar 07, 2003 11:23 am

I have two wave forms that are nearly identical, each with a few unique variants. They are approximately the same length in time, have the same sampling rate, etc, and I want to mix them together, but do so with the logical AND statement... so the result is only the waveform data that exists in both samples.

For example, any data in A that is not in B will not end up in the final waveform C, and visa versa.

I would like to specify a tolerance level as well, since they are not exactly the same-- give it a little room to loosly match the core waveform.

Before anyone flames me, please note that I've RTFMed as much as I could, and have searched through the forum and didn't find anything particularly relevant. (Perhaps I just don't know what verbage to search on...)

Any help, or leads in the right direction would be appreciated!

-=-=-=-=-=-=-=-=-=-=-=-
Don't Panic Tongue

VoodooRadio · Location: USA

Are you describing.. pasting and inverting? I'm a little foggy here. Shy

jonrose · Location: USA

Sounds to me like at least a two-part operation, perhaps three.

Maybe younglove or another CE scriptwriter here on the forum could help you out with this one.

Best... -Jon

RabbitDance · Posted - Fri Mar 07, 2003 3:00 pm

If I read your post correctly, it seems the solution you are looking for is similar to 'keep the vocals'(center channel), when a two channel waveform has center information common to the two. In a word 'no', you can't only keep the information common to the two and only that information.

There is something you could try and that would be the 'average' preset in the Amplify -> Channel Mixer effect. Combine the two waveforms into one two-channel by copying and pasting. Then apply the effect. This will in effect double the volume of anything common to the two, but will leave unique information at the same level. Please note, the above suggestion assumes the two waveforms are lined up exaclty and are in phase with eachother.

Inverting and pasting (L-R) could leave you without comon information, this is otherwise known as the 'vocal cut'. However this is the opposite of what you asked for.

Some clever filtering, either eq, or outright cutting or silencing might be the only thing that will work for you. This could be cumbersome but might be the only satisfactory solution.

If you are a programmer, you could loop through both waveforms and take the lesser (volume) of the two. This might roughly give you what you want but assumes 1) you have the programming skills 2) you don't mind lots of noise introduced.

That last one was a bit of a joke, I don't think the results would be satisfactory. Good Luck!!

Mark T · Location: Norway

I would think you could at least get started by doing a invert/paste which gives you the differences then invert/paste that back onto the original which should leave the commonality (or somewhere near it). I think you would need to experiment with invert/paste A to B then invert/paste the result to A (or B)? and you might have to do it twice (once to capture A-B and once for B-A).

Good luck anyway! Wink

alofoz · Location: Australia

Hmmmm.... you could certainly write a program to do what you describe, but I think the results would be awful.

zemlin · Location: USA

Finding what is different between the WAV files is easy. Paste/Invert file A onto file B will cause anything common to cancel.

If you take the results of a paste/invert of A+B and paste/invert that onto A, you should end up with A without the differences. (It's early in the AM, I think I have that sequence right) As far a tolerance, this technique will have none.

If A&B are not made of material from the same original recordings, you might be out of luck as timing and phasing will not match closely enough to allow safe mixing.

zemlin · Location: USA