Bit Depth (And Sample Rate Part 2)

This is part 2 of my previous post on sample rate. 

 The video below shows a full 16-bit Game sound sample. The difference from the previous samples is insane. You have vocals, guitars, drum kits. But notice ... it's just a little bit ... washy sounding.  

OK that was a trick ... if you have the quality set to 480p, the washiness should go away, mostly. But now that you are aware of it, listening to youtube music on low fi settings will now dive you mad. Sorry!

OK still with me? Here's the thing. Bit depth makes a huge difference. And 32-bit is even better than 24-bit ... 4 billion values of depth is nothing short of overkill, but it has it's advantages (more on that some other time).

But what about sample rate? I talked about it in my previous post, but I didn't get into the nitty gritty other than to show you that recording at 192khz can be redundant. (although up-sampling during mixing and mastering can be helpful).

Sample rate is essentially the number of times a sound is captured by the sound card each second (bit-depth is a combination of the lowest Digital loudness value that can be written, and the sheer number of depth values that can be written). It stands to reason that at 44.1khz (which is 44,100 hertz), a bit depth of 16-bit is fine since you can write a little over 20 thousand more values than the samples available. 44,100 samples available, 65,000 levels of depth available. The math works.There is a lot of buzz about recording at 24-bit, and I do record, mix and master at 24-bit. But I mostly do it for posterity and for the fact that even small volume changes can introduce many more bits to the audio signal. The cold hard fact is that I have fans, guitar amps and room ambiance in my recordings which create a noise floor that is above 16-bit.

But I've been using 24-bit for so long that I'm going to just keep doing it. Besides, 2-bit is the new standard for HQ audio, so if even for that I will continue to use it.  It doesn't seem to take up that much more processing power ayhow. On the other hand, upping the sample rate makes every plugin work harder because it has to process millions of more values every second.

Make sense? Hope so, cause I'm moving on!

The more times a sound card (Audio interface I mean ...) captures a sound, the more crisp  the sound will be to the human ear, to a point. Sample rate (just like human hearing) It's measured in something called Kilohertz (abbreviated KHz). A CD has a sample rate of 44.1 Khz, which is a little over twice the 'sampling rate' of the human ear, which tops out at 20 KHz. 

Now the first bizarre thing about digital sample-rate is that the very top frequency produced will be roughly half the value of the sample rate. So a 44.1khz sample rate has an absolute top value of around 22khz, which means you can do EQ adjustments up to 22khz but no more using the 44.1khz sample rate. A 48khz sample rate lets you do adjustments all the way up to 24Khz - and that's obviously not really necessary.

BUT

The second bizarre thing about digital sample rate is a thing called "aliasing".

Here is an awesome video that shows exactly how aliasing works. See my last post for a test to see whether your sound card/audio interface deals properly with aliasing and inter-modulation distortion.

So where the rubber meets the road on sample rates is that while a 44.1khz sample rate is good, if the anti-aliasing filter used by your RECORDING INTERFACE is poorly designed, you will hear this nasty aliasing in your recorded music! Ouch!

Plus, it will screw up your monitoring!! It would be all in all a very unfortunate thing. But it's a thing that I experienced with two pieces of gear. Not to knock M-audio (I did email them many times and got no response ... so hey, you get what's coming). The Firewire 18/14 and the Fast Track pro. When I recorded at 88.2khz, the sounds would be crisp and detailed. But when I recorded with the same microphones, the same instruments, the same settings and a minute later at 44.1khz, the aliasing would be unbearable (and shows up all over the place in my first album!). Since I had low computing power, my only solution was to record at 88.2khz, then down-sample the material myself. I did that with two songs that didn't appear on my first album, but that are slated for release on my second album. I used the original parts, so I'll be sure to load them up and do a breakdown once it's released!

So the lesson here is simply this: record at 44.1khz only if you are sure that your gear has a good anti-aliasing filter and no inter-modulation distortion at lower sample rates. I would like to assume that most modern gear deals well with these issues, but you never know! If you can afford the extra processing, make 48khz your go-to for recording, mixing and mastering. After all, 48khz is standard for DVD, and mastered for Itunes recommends sample rates of at least 44.1khz, but higher seems to be preferred. Also, if you release on vinyl, it's good to have masters that are 48khz. 

I would also do a little research and use audio interfaces known for their quality. You don't have to break the bank, either. I love my tiny but powerful Focusrite Scarlett 2i2. If you need the bells and whistles, M-audio makes sturdy gear, but you might have to play with it to make sure you aren't introducing nasty artifacts into your recordings if you work at sample rates below 88.2khz.

Here's to your music!

Ryan