There’s no real answer to that as it’s more about personal choice than a technical imperative, and depends so much on what/how you are recording and what the final resolution of the file will be. Of course, 24/96 theoretically gives you higher resolution audio so you may as well use it to optimise the results, but some would argue that most of us can’t hear the difference anyway.
Try recording the same source at 24/96 and 24/48 (original recordings of the same source, no sample rate conversion) and then see which file you prefer in a blind listening test. Let your ears make the choice.
I say no, stay at 24/48 and save unnecessary stress on your system. Most people who say they can hear a difference are lying themselves. But… If you can hear the difference, you should perhaps consider a career in mastering!
I record at 44.1 and at 24 bit. I can’t hear any difference above 44.1khz
Am I right in thinking that recording at higher sample rates gives less Latency? If so, that would be the only reason why I think it would be better to record above 44.1khz (unless its for Movies/Tv etc and then obviously at 48khz). I could be wrong though…
It’s not just the hard drive space. It’s also plug-in count. If you double your bit rate you effectively halve your plug-in count, which is not good if you like doing sound design and experimental sound effects.
It’s case of contention much argued, so try it if you must but, personally, I have seen no evidence of any difference of note so I stick with 24/44.1 and I reckon my productions sound just fine. If I wanted to spend money on pro mastering and what not, probably even better. There are some stunning tracks from the nineties recorded in 16/44.1 long before the advancement in today’s technology .
‘Bit depth’ (more actually ‘word length’) describes the dynamic range. It makes sense to record at 24 bits as this allows you to leave plenty of headroom to avoid clipping, while the noise floor can be very low. It makes sense to have all your projects at that word length.
Sample rate is a more complex topic. Nyquist theorem dictates that (with a perfect filter in your A-D converter) your sample rate must be at least twice the highest audible frequency. Most humans can barely hear above 21kHz. So 44.1kHz should be sufficient. Cheap and older converters require a bit extra to allow their filters to do the job, so 48kHz is probably better. Some sources which carry ultrasonic energy (cymbals, keys being shaked etc) can cause intermodulation distortion in the audible band. So it can make sense to track at 96kHz if you can hear problems. You can record at high sample rates ifvyo need to slow down/pitch shift down for sound design.
All of which leads me to track at 48kHz in normal circumstances!
There is a gazillion different views on this. Many Sound on Sound articles about this over many years. You can also research the enormous threads at Gearsluts and Pro Tool forums about this. Many mastering engineers also wrote many articles on the issue.
Everyone has an opinion on the issue & many schools of thought.
Mastering Engineers ( most of them ) all recommend working as high as you can, and they also among themselves differ on the reasons.
Personally I won’t go under 44.1 kHz / 24 Bit - but I work at the moment @ 96 kHz / 32 Bit float - and my mastering chappies are happies.
If you have the speed and space go as high as you can - if not go for 44.1/24 at least.
If you do work at 24bit - use the dithering plug (UV24HR) set to 16 bit in slot 8 master buss when you print your final mix for CD - if you master yourselves and your mix is done. (Not while you are mixing)
Well this is interesting because there is a frequency I hit on my sampled piano in a specific and narrow range (could be one key) that causes distortion which is audible on some speakers. I do not hear it on my studio monitors though. It is particularly apparent on cheap speakers.
I do sound design on occasion. As I understand it, a high bit depth isn’t necessary if your plugins do this for you. This is called “oversampling.” I am not aware if Cubase 8 plugins do this or not.
As for which sample rate and bit depth to use…
it really depends on the source you are capturing and the room you are capturing!
I wouldn’t bother with 96 khz on a distorted metal e-guitar recording but I would absolutely use it on an acoustic guitar/piano/drum/orchestral room mic recording in a beautiful sounding naturally reverberant room.
Realistically, the ONLY way to learn the difference (when to use it and if it matters at all to what you do) is to try recording a song at 96khz/24bit and mix it out to hear the end result.
(And anyone who says you won’t hear a difference has probably never tried it themselves or doesn’t record sources where it will even matter!)
You’ll also want to check the specs on the plugins you use and make sure they handle your choice.
WAVES publishes the specs on their plugin pages (as do most other deveolpers) for reference.
The majority of the recordings I do are 48khz 24bit or even 44.1khz 24 bit (and almost always mixed at 32bit float no matter what sample rate!)
And you’ll easily hear a difference between 24bit and 32 bit float if you do a lot of rendering or track to track recording (or a “Print track”) during your mixing process.
Think of it much like the difference between a picture taken on a 1.5megapixel camera and a 16megapixel camera … most people can clearly see the difference! The same holds for good ears… good ears will hear a difference.
But for audio the real difference in bit depth occurs during the mixing process!
Yes, if you have a sample buffer of a fixed size (let’s say 256 samples), you will find that you experience less latency at 48K than at 44.1K, since 48K processing burns through samples faster than 44.1K.
The problem is that this also means that you’re getting less protection from those 256 samples than you were getting at 44.1K.
On top of that, processing audio at higher sample rates demands more DSP/CPU than lower sample rates, meaning your likelihood of needing that buffering increases as the sample rate goes up, so you might find yourself having to increase the size of your buffer, which will increase the latency again.