soft sampler: record 16 or 24 bit?

[nakmon]

I'm a newbie, so please be gentle. Thanks in advance for any advice given.

Here's my question: As I understand it, most run-of-the-mill samples are recorded at 16 bit 44.1 khz. When using such samples, is there any advantage to recording at 24 bit when using a soft sampler? Or am I way off track (no pun intended).

This is assuming that everything stays digital, without sending the signal to outboard processing gear.

I'm thinking about Sonar and VSampler as a Dxi when I ask this question.

[drstawl]

Theoretically a 24 bit sample could sound a lot better than a 16 bit sample. But it really depends on the sample. A good 16 bit sample will trounce a shitty 24 bit sample any day of the week.

Other than that the only variable involved would be the extra processing overhead associated with the larger sample size- negligible on a late-model PC.

[nakmon]

Maybe my question was a "bit" confusing. If the samples are themselves 16 bit 44.1khz, what good does it do to record them in 24 bit? Again, assuming the soft sampler output is recorded internally with no D-A-D conversion.

[The Byre]

If you are using a hardware sampler, not a great deal. If you are using a software sampler, 16 bit will mean less data for the computer to process and therefore fewer problems. With a SW sampler, you will get latency problems that are greater with 24 bit.

Latency problems is the amount of time it takes from telling the computer to play a sound to the actual sound coming out of the apeakers or being recorded.

You can get round this by recording in MIDI only using a cheap shitty 8-bit sample to hear what you are playing and then rendering your sample off-line using a more complicated sample that would give you latency problems if you were to play live.

[dobro]

Mmm... I don't see much advantage to recording a 16-bit sample at 24-bit, but what about when you start mixing and tweaking, and all that dithering starts? What then? I'm asking, not suggesting.

[nakmon]

Thanks for the replies.

Hmm. I didn't think about increased latency from the computer's extra processing at 24 bit. Since I don't step record in MIDI, would this hamper laying down the MIDI tracks?

I should say that at this time, this is all theoretical. I currently use a hardware sequencer and record to a digital four track. I fooled a little with a friend's Sonar and soft synths/sampler and think it might do the trick. I intend to get an audio only computer in the near future.

MOstly, I'd sequence everything in MIDI using the included Roland GM sound canvas just to get the song down. I'd later replace them with better quality samples and render each track one-by-one to audio. Add plug-ins and mix down.

Dobro, if I understand you correctly, the more plug-ins used to treat the audio, the more degradation of the original sample? If true, is that a reason to go to 24 bit?

[TexRoadkill]

Dobro was hinting at the right reason. Recording in 24bit will give you more dynamic range for processing. This means that reverb and delay trails will be a little more distinct as they fade off. If you are going to be combining any live vocals or other instruments than it also gives you the ability to record them at the higher bit rate.

Having the entire project in 24bit also gives you more data to work with for any DSP and can help reduce cumulative errors. If you are not doing any internal processing then the gains would be minimal.