[arnyk]

  There are genuine advantages to recording and editing in 24 bit, specifically due to available headroom and combined noise floor (if you mix a bunch of 16 bit dithered files together). For playback, 24 bit offers no benefit though.

 
If you edit just one file or two files or whatever, 16 bits is more than enough. 24 bits allows you to be incredibly sloppy. Working with 16 bits should be well within the capabiliites of any professional recordist.

 

[Steve Eddy]

There are better formats than JPEG like BMP or TIFF which in a way are like the FLAC & ALAC of the image world.  [I was making a comparison].

I'm talking about for web use. But the example still holds.

se

 

[Steve Eddy]

If you edit just one file or two files or whatever, 16 bits is more than enough. 24 bits allows you to be incredibly sloppy. Working with 16 bits should be well within the capabiliites of any professional recordist.

One file? Two files? What about 16 or more tracks?

And incredibly sloppy? How much studio recording experience do you have, Arny?

se

 

[arnyk]

One file? Two files? What about 16 or more tracks?

And incredibly sloppy? How much studio recording experience do you have, Arny?

se

 
The point was already made that the more tracks you mix, the less you are dependent on their individual quality. That falls out of the arithmetic of mixing and like so many things, the math predicts the actual practice.
 
I have almost no studio mixing experience but at least 12 years of professional live venue recording, and mixing live sound.
 
Trust me, studio mixing is far easier than live sound. Every time I had an assignment of that nature, it was a walk in the park - such a controlled environment!

 

[interpolate]

Not being a prude although 16-bit and 24-bit are the correct terms.

 

[arnyk]

Not being a prude although 16-bit and 24-bit are the correct terms.

 
While that may be correct reality is that audio editing software generally works internally with 16 bit fixed point data and 32 bit floating point data.  Because it conforms to no particular computer's internal data flow, 24 bits generally exists only as an external file format, and data is usually converted to/from internal 32 bit floating point data.

 

[Hudson]

   
While that may be correct reality is that audio editing software generally works internally with 16 bit fixed point data and 32 bit floating point data.  Because it conforms to no particular computer's internal data flow, 24 bits generally exists only as an external file format, and data is usually converted to/from internal 32 bit floating point data.

 
Actually this is something I've always been curious about audio editing software. If you are using a computer, isn't all of the computational processing done using floating point arithmetic? I mean where is the delineation between floating point and fixed point arithmetic along the processing chain?

 

[arnyk]

   
Actually this is something I've always been curious about audio editing software. If you are using a computer, isn't all of the computational processing done using floating point arithmetic? I mean where is the delineation between floating point and fixed point arithmetic along the processing chain?

 
 
Depends on the software package.
 
The most general rule might be that all more complex operations are done in floating point, and converted back-and-forth to fixed point as needed when the file is 16 bits. 

 

[Hudson]

Yeah a quick google suggests that most pro audio processing tools utilise floating point processing.
If that is the case, I'm curious are there really any areas along the processing chain where high bit depth is of any benefit?
Providing of course you are of the opinion that the redbook standard is satisfactory for playback!

 

[jcx]

internal digital processing can have "surprises" - for those who haven't read literature on wordlength effects, coefficient quantization, rounding that dates back to long before computers, DSP was a consumer product possibility
 
in particular if you want to emulate analog continuous filters with IIR blocks - some internal accumulators have to be much longer than any in FIR filters when the filter corner frequency is a small fraction of the sample rate - 32 bits may not be enough
 
the last fixed point DSP I programmed was 16 bit wordlength and used 40 bit accumulators in the FIR, I needed to go to 48 bits to keep  a smooth 16 bit output in IIR filters with 10 Hz corner frequency and only 16 kHz sample rate in an Industrial instrument
 
most of the analysis is "just arithmetic" but some subtleties involve nonlinear effects if you truncate wordlengths wrong, have "biased rounding" it is possible to get very low level cycles in the noise - "birdies", noise modulation

 

[interpolate]

Interesting as this is, this topic is starting to meander. So have we agreed that hi-res audio is worth the extra money or not?

 

[nick_charles]

  Interesting as this is, this topic is starting to meander. So have we agreed that hi-res audio is worth the extra money or not?

 
No we have not agreed either way.
 
My personal take is that (1) I have not seen sufficient strong [the plural of anecdote is anecdotes not evidence] evidence that suggest that folks can generally tell competent high-res from competent non hi-res taken from identical masters and (2) since I can only rarely and generally with pathological cases distinguish between lossless and high bitrate lossy of the same source and (3) everything above 13Khz is an unknown land for me at my age, it is probably not worth it for me...
 
YMMV of course 

 

[Music Alchemist]

  Interesting as this is, this topic is starting to meander. So have we agreed that hi-res audio is worth the extra money or not?

 
It could be worth the money for you if a different master is used, in which case it would definitely sound different. Otherwise, if for example you already have the CD and the hi-res download is the same master, it will sound the same and you will just be buying the same thing twice. The problem is they don't normally disclose such information.

 

[arnyk]

  Interesting as this is, this topic is starting to meander. So have we agreed that hi-res audio is worth the extra money or not?

 
The classic high resolution proof of performance would be to compare the identical same music, recorded at the same levels and with the same mastering,,.processed using best practices,  recorded both ways, and a reliable audible difference would be found.
 
This is actually a pretty simple test, and it has been tried many different times, many different ways over the past 15 years.
 
Critics of so-called high resolution audio have consistently failed to hear this difference, and  so have its advocates.
 
One of the largest scale tests of high resolution audio happened in the years approximately years 2000-2008 when about 50% of all so-called high resolution commercial releases were sourced from low resolution masters. No high end audio reviewer, critic or audiophile seems to have noticed. At least there are no published reports of this.
 
Eventually critics of high resolution audio noticed this problem, based on technical tests.

 

[Music Alchemist]

  The classic high resolution proof of performance would be to compare the identical same music, recorded at the same levels and with the same mastering,,.processed using best practices,  recorded both ways, and a reliable audible difference would be found.
 
This is actually a pretty simple test, and it has been tried many different times, many different ways over the past 15 years.

 
Yep! All you have to do is convert the "HD" files to 16-bit / 44.1 kHz to find that you really can't hear a difference between them.