A question about bitrates and needledrops.

when people say cds are 16/44.1 are they talking about the file size and how it fits on a cd or are they are saying that this is what a cd player can only read?
by that i mean, if i was to burn my needledrops to cd using 24/44.1 or 24/48, would any cd player be able to read the cd? and if so, would it have the edge in sound over the standard 16/44.1 cd?

 

Hi dj2hynes,

16/44.1 is the format for CD audio. It is the only format a CD player will read.

The 16 refers to how many bits are in the digital "word".
The 44.1 refers to how many digital "words" (or samples) there are for every second of audio.

Using longer words, like 24-bit words will, to my ears, most definitely "have the edge in sound over the standard 16/44.1 CD" but you will not be able to play these in a CD player.

The word length will have to be shorted to 16 bits. If you simply convert the 24-bit file to 16-bits, the bottom 8 bits will get chopped off ("truncated"), taking with them a lot of fine detail.

The best way to shorten a long word length file is to use a good dithering algorithm. This will preserve some of the information from the bottom 8 bits when the file is converted from 24-bits to 16-bits. Dithering should be performed as the very last step before burning the file to CD.

Unfortunately, most of the dithering algorithms included with most software will add their own colorations to the final 16-bit file. If you are simply doing needledrops, with no changes after the file is digitized (including level changes), the easiest way to proceed is to do the needledrops at 16/44.1.

If you want to make changes, like level or EQ changes, I would suggest doing the needledrops at 24-bits, making your changes and last of all, dithering. Then you can burn the dithered file(s) to CD-R.

Hope this helps.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

 

Barry, isn't the general rule of thumb to record and process in 32 bit to avoid rounding errors then do the final dither down to 16? That is, unless you're doing no processing at all. In that case recording straight 16 bit is fine.

 

Hi Barry,

How much "change" in the sound have you noticed with doing some level changes with a 16/44.1 file? Is it very, very slight or obvious?

Best,
Todd

 

Just a slight clarification...

44.1kHz = 44100 samples per second ( and x 2 for stereo).

I know you know this, but just pointed it out in case someone didn't realize it.

 

1441 kbps.

 

Some CD players do not read burned CD discs at all. Some of your friends having such ones, may be disappointed.

 

I accidentally burned a CD-ROM from a 48K (DAT) file.
The player played it but it ran very slow.

 

How old are their CD players? I haven't seen one made recently that wouldn't.

Ed

 

Hi alethos,

What you say is true for some applications and not true for others.
Since the original poster asked about 24-bit vs. 16-bit, that is where I focused my reply.

Keep in mind, so-called 32-bit is much like 24-bit but with the ability to "slide" up and down. (That may not be the best explanation but it is what comes to mind without resorting to mathematical terminology.) In other words, it is not "8-bits better" than 24-bit in the same way 24 is over 16.

Some applications process their temporary files at the word length of the source file. With these, one can get slightly better results by encoding at 32-bits (or 32-bit float). The very best applications will process internally at more than 32 (e.g. 48-bit with some parts of the "engine" working at 64-bit) and work with 24-bit source files.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

 

Well, I would have said: "Some of your friends "vintage equipment lovers" or some of your relations "80's-90's good old time machines owners" may be disappointed"... OK?

 

Hi Todd,

The answer to your question may well depend on the listener.
In my work, I would never make the slightest change to a 16/44 file as the result is quite obvious to me - not at all "slight" much less "very, very slight".

What I hear is that low level information, where the fine detail, the subtle instrumental color and just about all of the sound of the room and any reverb in the recording, is simply eradicated.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

 

Barry

In extracting audio files from CD (16/44) and then performing volume changes, edits and fades, I can't say I've ever noticed anything as drastic as what you mention here.

 

That has also been my experience.

 

Hi Barry,

Should I understand from this that iZotope's 64-bit SRC will deliver the same results with a 24-bit file than with this file changed to 32 bits, saved, and then reloaded?

 

Hi Paul,

Like everything else in audio, much depends on the listener, what they're listening on, how it is set up, etc.

On my system and most of the ones I listen to (even the little one in the kitchen), what I hear are drastic (as you put it) losses that I would never allow in work folks pay me to do or in work I do for myself.

If you don't hear anything that bothers you, you can certainly save yourself the extra steps that would be required to avoid it. In my experience, all folks hear things differently.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

 

Hi Metoo,

iZotope's SRC is only part of the equation. The rest is in the software that you use to host the SRC. It is in the host software that the requirement for 32-bit files for processing sometimes comes into play.

What are you using the iZotope SRC in? I would submit that some of the benefits of the iZotope algorithm are lost with a lot of host software.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

 

Right, I understand that.

I didn't realize it was application dependent. It makes sense if you stop to think about it...I just never had. Thanks for the info.

 

Well, you have introduced another viriable in the equation.

Just to leave the host software out of the equation (because I understand what you are saying, but the great variety of wave file editing software out there would dilute my question) let's just use - say - iZotope's own RX Advanced, which includes their SRC and MBIT+ dithering in its own program. Would there be any difference in this context between using 24-bit files, or the same converted to 32-bit float for, say, downsampling or Gain changes?

 

To avoid things getting complicated for the uninitiated I feel it is worth mentioning a few things:

1) I use audacity for needle drops at present. I prefer to record 32bit/96kHz then do the processing and downsample to 16bit/44.1kHz using the triangle dither option. I personally think this sounds nicer than straight to 16/44.1

2) I think it is fair and respectful to say Barry has greater hearing (or perhaps it may be more accurate to say greater awareness of what he is hearing) than the average bear.

With this in mind to the OP: experiment & if you don't notice a difference don't worry too much; make sure the end file is 16bit/44.1kHz if you plan to burn to audio format CD.

 

Hi Metoo,

First, I want to quote Yogi Berra (and this should be remember frequently in audio conversations):
"In theory, there is no difference between theory and practice. In practice, there is."

So, theoretically, IF you could use iZotope's RX or any iteration of their SRC or MBIT+ dither/noise shaping without a host, in my view, you could safely feed them 24-bit files and not suffer any loss at all.

Now, back in the real world, you're going to have to use a host application and that is going to determine how to get the best when you use that application and incorporate iZotope's wonderful algorithms in the process. With one of the apps I use, which processes internally at 48-bit (and partially at 64-bit), I'd feed the whole thing a 24-bit file. With some other apps, I'd feed the process a 32-bit float file.

So to answer your question, if not perhaps in exactly the way it was framed, with some apps, it won't make a difference and with others it will.

For those who might be new to this, I would add that I would not lose much sleep if I had to process a 24-bit file in any app. As I said in another post in this thread, 32-bits does not represent an additional 8 bits of data in the way 24-bits does over 16-bits. Much more important, to my ears, is the integrity of the dither/noise shaping algorithm that is going to be used later to create a 16-bit version for CD.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

 

Hi Mike the Fish,

If you are going to do processing (or even the slightest level change), I would highly recommend what you are doing, i.e. digitize and process at a long word length (and high sample rate, if possible).

To be clear, you are not recording at 32-bits (unless you have an A-D converter of which I've not heard). You are recording at 24-bits and having Audacity Save As 32-bits. There are no 32-bit converters (of which I'm aware) because what we call 32-bits not 32-bits. It is really a "fancy" version of 24-bits.

If you were sure you did not want to process or change level at all, I would suggest digitizing at 16/44 unless you have:
1. a converter you are positive does 24/96 better than it does 16/44 (many don't)
2. an application that can process "bit clean" 24-bit files (most can't, regardless of what the specs say)
3. a dither/noise shaping algorithm that will not exact a sonic price greater than what it gives in return (most do)
4. a sample rate conversion algorithm that will not add spurious harmonics that add an artificial sheen to the sound (most do).

Even if you do process and/or change levels, there are many degrees to which most applications and algorithms achieve (or do not achieve) what the four points above require to get the very best results. However, even within their limitations, the results may still be pleasing, depending on what you want.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

 

Thanks Barry.

 

You've lost me and we're diverting the thread. My question was innocent and I meant no harm. Nevermind...

Back to your regularly-scheduled thread.

Ed

 

Ok, I get your point. Thanks, Barry.