I'm trying to determine the best possible sample rate for my projects. Most of them are a combination of sample libraries and my own recording.
These are the two basic options:
96 kHz project setting
- I can record at 96 kHz (which my A/D converter can handle) to have the best possible recording quality
- As most sample libraries are recorded at 44.1 kHz, K2 and VST instruments will have to resample to 96 kHz
- I have to convert my final mix to 44.1/48 kHz for delivery / CD
44.1 kHz project setting
- I can only record at 44.1 kHz
- As most sample libraries are recorded at 44.1 kHz, there's no resampling going on
- I don't have to convert my final mix
Which of these would be best?
I remember Spectrum saying here on NS that digital sample rate conversion sounds bad. Whether you're going from 48 to 44.1, 96 to 44.1 or even 88.2 to 44.1. Even when using a quality converter like R8brain Pro?
Does this also apply to the resampling of your VST instruments going on inside your sequencer? Say I'm working at 96 kHz. Since most sample libraries are recorded at 44.1 kHz, K2 and VST instruments will have to resample to 96 kHz in Cubase. Will this also affect audio quality and would it therefore be better to pick a sample rate that the majority of sample libraries used in the project were recorded at, i.e. 44.1. And then also record at 44.1?
I am using 44khz after reading many threads about the topic. I think we have the same situation since I too use mostly vsts that are not 96 natively so upsampling them just to downsample again for the CD 16/44 makes little sense to me.
Maybe someone can add to this but my projects are now 24bit/44khz
Most importantly, I think you need to figure out what your final output needs to be, then work to that. If its Cd - 44.1, stick to that, if its 48, work at 48. I find very little reason to work at 96, takes up too much resources for me. There have been times when I needed to export 96k files, in which I worked at 48 and upsampled the final file to 96, sounded fine. upsampling, there will be no improvment in quality, only adding information and data to a file, but no improvement, or likewise I haven't noticed any bad effects either. Working from samples that are 96k , then dithering down to 44.1 for your final output, eh, you may notice just a small reduction in quality. For me, bit depth makes more of a difference, 16-24, rather then sample rates higher then 44.1 do to me. I can notice when a file goes from 24, to 16 easier.
If you work with libraries that are at 44.1 - 44.8, and record at 96, you shouldn't have a problem, as least I didn't when I have, it will bog your puter down quite a bit more, so just make sure there is a reason you are working at that higher rates.
In that case, you can make great sounding mixes at 44.1 or 48, a higher sample rate doesn't necesarrily equate to better final audio. Some systems operate differently, where people say there setup just sounds better at 48, or 44.1, I guess different converters and audio cards might work better at different setups. There use to be something to keeping your sample rates even for better up and downsampling, such as if you are going up to 96, record at 48, evenly divisible numbers, however with the improvement of algorithms and converters, I don't believe this is the case anymore.
Actually, I want to do my own recording at 96 and most of my sample libraries are 44.1. So I'm trying to come up with a definitive workflow here.
I was mainly wondering how bad the realtime resampling of your VST instrument outputs by your sequencer is (Cubase SX3 in my case). If all your sample library content is 44.1 and you set your project to 48 or 96, all this audio will be upsampled in realtime by Cubase. In that case wouldn't it be better to work at 44.1 and don't upsample to 48 until at the end?
If you work at 44.1, then upsample to 48, you are doing nothing but adding extra 1's and 0's to your files, making them bigger yet not improving them one bit. Is there a reason you want to work at 96k? Are you going to be recording additional acoustic material along side your sample libraries? If you are just working with sample libraries, and they are recorded at 44.1, or 48, then there is no reason to work at 96 because you will not be gaining anything one bit. the samples are already recorded, the frequencies are already captured, makes sence?
If you want to work at 96 because you want to record additional stuff where the higher sample rate will make a difference, then you shouldn't have a problem with recording in your samples at the higher rate, just know it won't be improving them one bit. Also, keep in mind that unless you are going to be exporting a 96k session for a particuiliar reason, there is really no popular medium to play it back, so if you want to hear your stuff in a cd player, you will have to yet again dither down to 44.1, right back to where you started.