Dithering about dithering. And what order when changing sample rate & bits p/ sample?

[Max Dread]

Hi all

I've just finished sorting through a bunch of downloaded and random albums (FLAC and mp3) ready to tag and integrate into my library of ripped CDs etc.

As part of that process I like to convert anything with a different sample rate or bits per sample to 44.1khz/16-bit.

Dithering. I don't remember there being the "Apply Dither" option in the Bit Depth DSP. I've left it at none. What's the best option? I've got back ups so could redo them with dithering if it would be better...

Also, when converting sample rate and bits per sample at the same time, does it matter what order the two DSPs are loaded/added? I memory serves, the DSPs are actioned sequentially, so I wasn't sure whether they'd be better off a certain way round?

Many thanks

[Spoon]

Re: Dithering about dithering. And what order when changing sample rate & bits p/ sam

You can do, it does not hurt.

Bit depth - floating point
Resample
Bit depth 16 bit triangular

[Max Dread]

Re: Dithering about dithering. And what order when changing sample rate & bits p/ sam

Thanks for the reply.

Good news that it doesn't hurt to convert. I don't like pain When I A/B tracks before and after converting, I've not heard any differences to date, so that's good.

"Bit depth - floating point". I'm not sure what you mean or what that is or what you're referring to? Sorry.

"Bit depth 16 bit triangular". I've already converted using "none". How much difference does it make and is it enough to warrant going back and redoing them? Is it all about the sound quality or are there other factors it has a bearing on?

"The correct order". Should it be as you've listed, and if so how important is that? Tbh, I didn't give it a second thought until after I'd finished so I don't even know what order I had the DSPs in. Again, does it warrant going back and redoing them with the correct order?

Many thanks

[Spoon]

Re: Dithering about dithering. And what order when changing sample rate & bits p/ sam

You are probably not going to hear any differences between dither and no dither.

[Max Dread]

Re: Dithering about dithering. And what order when changing sample rate & bits p/ sam

That's good news then, thanks for letting me know.

And does the order of the DSPs make a difference? Bit depth change then resample? Or resample and then bit depth change?

Cheers

[Spoon]

Re: Dithering about dithering. And what order when changing sample rate & bits p/ sam

Yes, they are executed in order.

[Max Dread]

Re: Dithering about dithering. And what order when changing sample rate & bits p/ sam

Yes, they are executed in order.

Thanks, but what's the best order generally speaking when changing both?

[Spoon]

Re: Dithering about dithering. And what order when changing sample rate & bits p/ sam

Tbe exact order I gave before.

[Treelady]

Re: Dithering about dithering. And what order when changing sample rate & bits p/ sam

Thanks, but what's the best order generally speaking when changing both?

Let me take a stab at this.

When executing these jobs, do all sample rate conversions first. However, if your source file is not 32-bit floating (odds are they are fixed, not floating), then you'll want to do a the prep-process of changing the file to a floating point bit depth. (Some SRC utilities do this automatically behind the scenes. dBPowerAmp does not. I am assuming that is because the design philosophy is to do nothing to your files unless you specifically want something to be done, which is a very reasonable position).

HERE ARE THE STEPS / Order and settings of your DSP operations

(1) BIT DEPTH - Radio button "fixed" the drop down box choice should be set to "32-bit float" ALSO drop down box for Apply Dither should be set to "(none)"
(2) RESAMPLE - Radio button "Resample to Frequency" for the drop down box [select the sample rate you want the final files to be]
(3) BIT DEPTH - Radio button "fixed" the drop down box choice should be set to [the bit depth you want the final files to be], ALSO drop down box for Apply Dither should be set to "Triangular (TPDF)"

So, those are the three DSP processes, along with the correct settings for each one.

Notes
A) By changing the bit depth to 32-bit float, you give the sample rate conversion a floating point source, which, for our purposes, gives allows values to expand as far as they need to should there during the sample rate calculations. For example, you have to take a math test. The proctor gives you a letter sized piece of paper for all of the work you need to do for the entire exam. That's fixed point. Floating point would be that you're allowed to have as much paper as you need to do your calculations. An oversimplification, but hopefully that helps understand why this is step 1.
B) Then, do the sample rate conversion. That's pretty straightforward as to why. It was one of the things you wanted to do to begin with.
C) Dither - assuming you're going reduce the bit depth to something lower than 32-bit float, you want to dither. Otherwise the files are truncated (think digital guillotine) rather than being processed in a way that convinces human ears that the content close to the original source material.
D) I'm writing these in layperson's terms. There are more specific reasons / explanations of what each of these processes do, how the are calculated, and why you would choose one option over another. That's fine and well, but all you really want to know is 'What is the order to do this stuff, and if I'm in there, should I bother to dither?"


I hope this helps.
GH

[Everett]

Re: Dithering about dithering. And what order when changing sample rate & bits p/ sam



Let me take a stab at this.

When executing these jobs, do all sample rate conversions first. However, if your source file is not 32-bit floating (odds are they are fixed, not floating), then you'll want to do a the prep-process of changing the file to a floating point bit depth. (Some SRC utilities do this automatically behind the scenes. dBPowerAmp does not. I am assuming that is because the design philosophy is to do nothing to your files unless you specifically want something to be done, which is a very reasonable position).

HERE ARE THE STEPS / Order and settings of your DSP operations

(1) BIT DEPTH - Radio button "fixed" the drop down box choice should be set to "32-bit float" ALSO drop down box for Apply Dither should be set to "(none)"
(2) RESAMPLE - Radio button "Resample to Frequency" for the drop down box [select the sample rate you want the final files to be]
(3) BIT DEPTH - Radio button "fixed" the drop down box choice should be set to [the bit depth you want the final files to be], ALSO drop down box for Apply Dither should be set to "Triangular (TPDF)"

So, those are the three DSP processes, along with the correct settings for each one.

Notes
A) By changing the bit depth to 32-bit float, you give the sample rate conversion a floating point source, which, for our purposes, gives allows values to expand as far as they need to should there during the sample rate calculations. For example, you have to take a math test. The proctor gives you a letter sized piece of paper for all of the work you need to do for the entire exam. That's fixed point. Floating point would be that you're allowed to have as much paper as you need to do your calculations. An oversimplification, but hopefully that helps understand why this is step 1.
B) Then, do the sample rate conversion. That's pretty straightforward as to why. It was one of the things you wanted to do to begin with.
C) Dither - assuming you're going reduce the bit depth to something lower than 32-bit float, you want to dither. Otherwise the files are truncated (think digital guillotine) rather than being processed in a way that convinces human ears that the content close to the original source material.
D) I'm writing these in layperson's terms. There are more specific reasons / explanations of what each of these processes do, how the are calculated, and why you would choose one option over another. That's fine and well, but all you really want to know is 'What is the order to do this stuff, and if I'm in there, should I bother to dither?"


I hope this helps.
GH

Using Poweramp, I set the settings for converting the Bit Depth to 32-Bit Float from my Flac file which is set to 24Bit, when I hit convert I got this error message:

"Error FLAC accepts only PCM audio source. (BeginConversion)."

What am I doing wrong and how do I fix that error?

Oh - everything seemed to work when I just used the 16 Bit setting with Triangular dither and resample together. I'm really a rank amateur in audio processing, but I have a bunch of music (FLAC format) files with all manner of settings that need to be setup so people can burn them to CD if they want to.

Thanks in advance!
Everett

[Spoon]

FLAC cannot accept 32 bit floating point audio

[Everett]

FLAC cannot accept 32 bit floating point audio

, If I convert FLAC to WAV first, will that work?

[Spoon]

No, it will be floating point in wave, then when convert to flac same error, remove the DSP effect or change to non-floating point

[Everett]

No, it will be floating point in wave, then when convert to flac same error, remove the DSP effect or change to non-floating point

• Here’s what seems to have worked pretty well - is there some reason it somehow changes something I don’t know about? I Burned a couple of processed files successfully to Disk. I don’t have a CD Player - only what’s on my computer though, and the CD played under Quicktime and Music.

For 24BIT 48KHz to 16Bit 44.1KHz:

1.FLAC>WAV in dbPoweramp

2.Exit and open WAV in dbPoweramp

3.Change bit depth to 32BIT Float - No Dither

4.Exit and reopen dbPoweramp

A.Resample from 48 to 44.1
B.Change Bit Depth to 16 and under “Apply Dither” choose Triangular
C.Exit dbPoweramp

5. Burned cd