Subject:16-bit vs. 24-bit

Posted by: John 17:17

Date:3/17/2005 7:03:14 PM

Can some kind soul please help me understand what are the major benefits to saving my recorded .wav files as 24 bit instead of 16 bit? I believe my Audiophile 2496 soundcard is capable of recording in 24 bit, so should I always use that from now on? Is this something I should get in the habit of doing? Can the normal human ear discern the difference between 24 and 16? Does it make any difference between 24 and 16 when I burn the files onto CD? Or if I convert them to mp3? What are the differences, and when should I use each one? TIA

Subject:RE: 16-bit vs. 24-bit

Reply by: Lanco

Date:3/17/2005 10:52:51 PM

Hi John

I do know that if you are going to do any editing, like restoration, it's better to record at 24/96, do the editing there and then bring it back down to 16.

I believe Scott Garigus says that unless you edit there's no point in recording at 24 bit unless editing. That if you take a 16 bit recording it won't sound any better if recorded at 2496 and then lowered.

But, then again I remember a review a few years ago. I believe it may have even been for SF 5.0 which was their first 2496 program. At any rate, the guy posed that same question and did an A/B comparison and said yes the one recorded at 2496 and lowered to 16 sounded better than the one just recorded straight 16.

In a sense this makes sense to me that it would sound a bit (no pun intended) better because there is definitely more information recorded at the higher rate than is possible in 16.

It's interesting that you bring this up now, because I was going to attempt a little search on this subject like tomorrow.

But, mainly, yesterday I decided to record from my Yamaha Tyros arranger keyboard. I used just one shot clip. About a minute's worth.

I wanted to not only see what kind of difference, if any, the different rates made. I also recorded it with and without a Behringer UB802 mixer to see if the mixer actually would distort the sound like some people were telling me a few years ago.

On top of that I recorded the same tune at all the different rates that were available in Magix Audio Cleaning Lab, Cool Edit SE which I got with the Mia sound card, Audigy and SF5.0.

I have not yet had time to listen to them but for a few minutes. The first interesting thing I noticed was that the Audigy program sounded a bit livlier. I am not sure if it's just a little higher volume setting on that program or what. I think that's all it is.

I'll tell what I expect to find though. After really taking some time, I would be willing to bet that they all sound the same at their respective rates, in whatever they can handle. In other words, I expect all 16's to sound the same, etc. Not sure what the deal will be with 24 vs 32. The Mia doesn't support 32 as a statement in the knowledge base states.

Anyway, didn't mean to ramble. After I get the chance, I'll probably report what my findings were.

It will be interesting to hear anyone else's experiences on this.

Best

Scott

Subject:RE: 16-bit vs. 24-bit

Reply by: Rednroll

Date:3/18/2005 1:18:53 AM

Here's how it works. When you are recording audio and converting it from analog to digital, what you are doing is recording one sample at a time. Think of it this way. You put a sinewave on a piece of graph paper and you want to put a dot everywhere the sinewave crosses a point on an X-Y axis. The X axis is the spacing of the grid and this corresponds to the sampling rate. So let's say you have 1 second of audio that you want to break up into individual samples. If you have a sampling rate of 44.1Khz, so you have 44,100 X axis grid lines in that 1 second duration. Now for the Y axis, this represents the amplitude of the sinewave. The absolute maximun level is always 0dBFS. This is always the max sample value that can be represented on you graph paper. The number of bits represents the spacing of the grid lines on your Y axis. So if you are recording at 24bits you have 2^24 grid lines divided between a level of -inf to 0dB full scale. So 16,777,216 grid space values to choose from to represent that actual point the amplitude of the audio is when it is time to take a snapshot sample of that analog signal. For a 16bit recording you get 2^16 grid lines representing your Y axis or 65,536 grid lines divided into the same area on the Y axis. So if you're recording an analog signal you get a much more accurate sample representing the amplitude value because the grid lines are spaced much closer together as compared to a 16 bit sample grid.

Is 24bit's much better than 16bits? Well that's highly debatible wether you can tell the difference or not. Most will say you can definately tell a huge difference. I say, I listen to a lot of music and a lot of it was recorded before 24bit audio existed and others are more recent 24 bit recordings. If someone gave me a blindfold and told me i had to listen to a recording and I had to tell them if it was recorded at 24bit or 16bit, I think I would have a pretty hard time being able to tell. For editing Sound Forge does all your edits and processing at 32bit floating point, so you're really not going back to 16 bits until you save the file after you're done with all the edits anyways. Basically, 24bit recording is a more accurate representation of the orignal analog signal when you play it back. It's basically 48 times better representaton than a 16 bit recorded signal. One key factor when recording in 16bit is to make sure you record your audio signal peaking somewhere between -6dBFS to 0dBFS, to make sure you're using the maximum bit resolution that's available. 24 bit allows you to be a little more sloppy in having to maintain a high signal level to get a high bit resolution recording.

Subject:RE: 16-bit vs. 24-bit

Reply by: Phil Sayer

Date:3/18/2005 6:20:24 AM

Well put - the best explanation I've read on this subject - many thanks.

I record only spoken word - and have always stayed with 16-bit. I've now acquired a second-hand Yamaha 01V mixer (alas, NOT the new 96-bit version!) and I record using analogue into the sound card. (Yes, I know.... I'll do it one day! )

But if I allow the sound card and SF to run at 24-bit (which it will) how does this affect the WAV file? WIll my customers still be able to use the WAVs I send them without changing anything at their end?

Subject:RE: 16-bit vs. 24-bit

Reply by: planders

Date:3/18/2005 6:55:30 AM

The WAV file will be much larger, for one thing. You'll also find that not every computer (or sound player program, for that matter) can play back 24-bit files--don't count on the client having even an Audigy.

Subject:RE: 16-bit vs. 24-bit

Reply by: mpd

Date:3/18/2005 8:22:30 AM

One additional thing to keep in mind is that A/Ds are notorious for not being linear over the whole range, particularly for nearly full scale input. Recording at 24-bit give you some more headroom so you can back off the input gain so you stay in the converter's linear range.

Subject:RE: 16-bit vs. 24-bit

Reply by: John 17:17

Date:3/18/2005 11:38:59 PM

What about burning to CD? I'm totally ignorant about bit rates; can CDs be burned with 24 bit audio? My guess is yes.

Subject:RE: 16-bit vs. 24-bit

Reply by: planders

Date:3/19/2005 8:04:30 AM

Sorry, CD audio is only 16-bit, 44.1 kHz. You'd have to go DVD-Audio for 24-bit playback.

Of course, 24-bit WAV files can be burned to a data CD and played back on compatible computers, though.