Here's a protip for life in general - worry about the things that make a difference, don't worry about things that don't. Another - You'll die if you worry, and die if you don't. So just get on with life and stop worrying.
We get a lot of inquiries here at RP Towers from people with worries. The usual worry is that they don't have the right equipment. Sometimes they are justified - plugging a microphone into an interface with only line inputs is never going to lead to success. But no-one knows these things automatically so we are happy to help.
But quite often we find that people are worrying about things that don't really make any difference. We take it that anyone who follows RecordProducer.com and Audio Masterclass is primarily interested in making music and/or recordings of a professional standard. And by 'professional standard' we mean a quality that will satisfy a typical industry client, or sell into the market.
So let's have a look at the 16-bit to 24-bit conversion issue. Is it an issue at all really?
The first thing to consider is the significance of each of the bits. Those sixteen bits are used to quantify the voltage level of an analog signal in digital terms, in 65,536 steps (2 to the power 16).
So if at any instant the digital signal is at 1001010001101011 then that is a digital description of the instantaneous level of an analog voltage. In digital terms, the left-most 1 or 0 is the most significant bit. The right-most 1 or 0 is the least significant bit. Going from left to right, the signal is described with greater and greater accuracy, or you could say in finer and finer detail.
Sixteen bits are sufficient to describe a signal with a dynamic range of 96 decibels from its loudest to its quietest parts. (In real life it's a little less, but we'll stick with the theory.)
96 decibels is a HUGE dynamic range. Play some music really loud through your monitors. Now reduce the level by 96 dB. What can you hear? Nothing. Or as near to nothing as makes hardly any difference.
When a 16-bit signal is imported into a 24-bit session, all that happens is that eight more bits are added to the right of the previously least significant bit. In this case, they should all be zeroes as the 16-bit signal has no information here.
But what about dither?
Ah, there just has to be a complication....
One problem with digital audio is quantization distortion at very low levels. It sounds nasty and it is something to be concerned about. The way it is dealt with is to add a little dither noise right at the end of the recording chain. This eliminates the distortion and although adding noise sounds like a bad thing, it isn't - it's totally good. You will almost certainly have a plug-in for it.
So what difference would it make whether or not your 16-bit files were dithered?
Well if they are undithered what will happen is that quantization distortion will remain in the lower levels of the 16-bit signal.
If your 16-bit files are dithered, then all should be fine. Unless of course your DAW's import function adds dither without you knowing about it, in which case you have dithered twice so your signal is now noisier than it ought to have been.
The question is of course, do you know whether your 16-bit files are dithered? If they sound free from distortion at very low levels, they probably are. And if you can't hear any quantization distortion, well it doesn't really make much difference.
Dither is something that people can often get really worried about. Ideally you wouldn't add any yourself at all and leave it to the mastering engineer.
But why not explore dither for yourself - play around with some very low level audio -70 or -80 and below. Switch dither in and out; try different types of dither. When you hear what it does for yourself, you will be in a good position to make decisions.
In summary, if you need to import 16-bit files into a 24-bit session, just go ahead and do it. Listen closely - if they sound like they need dither then add it (at the 16-bit level). If they don't, then just leave them be. Why worry?