Is there a better/less noisy dither plugin than standard(embedded) available for Foobar to do 24/96 to CD conversion?
Serious? It will be hard to find a less noisy shape when it comes to audibility.
If you enjoy your music purely by watching spectral pics the situation may be different.
AFAIK the shape of the noise is a finetuned version of Naoki Shibata ATH based dither he introduced with SSRC.
Ok, so it seems that using standard filter is the only option ...
You may also use foobar, convert to 32bit and dither with another software but this i guess is of no help to easy convert.
Yes, of no help ...
I was just wondering whether noise shapind done by Foobar is appropriate e.g. versus plain TPDF as recommended by some sites.
Most recordings should not get near a 16bit noisefloor so everything but TPDF dither is overkill.
The idea behind using the shaped dither by default is to be always on the safe side.
If there really is extreme silence the shape will give you many dB headroom.
If the recording is noisy by its own you will hear the noise in the lower frequencies of the source long before the dither HF noise you spot by eye.
It works.
Most of nowadays recordings are effectively at ca 10-12 bit due to loudness war mastering. No matter is this is rock, techno, trance or whatever. Only classical music, some folk / national music recordings and other non-mainstream, quiet recordings can reach true 16 bits. So of course it always depends on what you are listening to if there is a reason to change such quirks.
Most recordings should not get near a 16bit noisefloor so everything but TPDF dither is overkill.
The idea behind using the shaped dither by default is to be always on the safe side.
If there really is extreme silence the shape will give you many dB headroom.
If the recording is noisy by its own you will hear the noise in the lower frequencies of the source long before the dither HF noise you spot by eye.
It works.
So why Foobar does not offer TFPD dither too?
Just wondering ... I know that probably i am puzzling with something not neccessary to puzzle with ..,
If you can demonstrate an audible problem you might have a leg to stand on. At the present moment I don't think there's any reason to overturn the status quo.
Well, I am OK with it, I was just exploring possibilities...
By the way, the issues that are not heard by general public can be heard/perceived under some special condition - http://www.aes.org/e-lib/browse.cfm?elib=17497
Jan
Been there done that (https://hydrogenaud.io/index.php/topic,107124.0.html); only interesting in that it raises more questions than it provides answers.
...but in a thread requesting TPDF, no less. Don't worry, the irony has not been lost on me.
I recommend you take that flimsy paper with a grain of salt.
The paper nowhere claims problems with noise shaped dither when i remember right.
OK, thank you for an opinion ....
That wasn't an opinion. ;)
...oh, wait you mean mine. Sure, no problem.
Do catch my edit regarding the irony.
OK, and I also agree than in thread about TPDF and other dithering no general discussion about can be heard/cannot be heard is appropriate,
To conclude, when converting 24/96 to CD standard 16/44.1 according to current knowledge foobars db/SSRC resampler and standard dither is enough, right ?
The paper nowhere claims problems with noise shaped dither when i remember right.
Except as a fuckup on their part, that paper had nothing to do with dither.
To conclude, when converting 24/96 to CD standard 16/44.1 according to current knowledge foobars db/SSRC resampler and standard dither is enough, right ?
Until you can provide evidence to the contrary, it should be enough; correct.
Thank you .... I have limited experience and wanted to be sure that when converting some FLAC 24 tracks to play on my cd player I am not losing quality more than neccessary.
The more relevant question for me is why they sell 24bit material as such when downsampled does not profit of noise-shaping ;)
How much of that material was originally 16-bit to begin with?
I know that it is hard to obtain true 24/96 recordings, I have only limited number of those tracks and majority of them are from my old vinyl sources which are pretty overplayed now. But this is not the main topic of this discussion ...
Provided they're peak-normalized, you'll probably be able to get away with truncating your old vinyl sources to 12 or 13 bits with no audible degradation.
I have found good TPDF dither DSP here http://www.saunalahti.fi/~cse/foobar2000/?changes=foo_dsp_dither
Of course it must be the last in converting chain and fobars dither must be disabled.
Adds "just white noise", useful e.g. when sound card aliases high frequencies and generally when noise shaped dither is not desired.
"Useful" based on first-hand discovery of real-world evidence, no doubt.
Where is the rolling eyes into the back of my head emoticon when I need it?
Alright, maybe I am not as advanced audiophile as the others here, so my posts could be searching for knowledge rather than proving it.
Anyway I tried the SoX command line program and find its dither options promising.
Also it is discussed (http://www.audiocheck.net/audiotests_aliasing.php) that some cheaper (but common) sound card can exhibit playback aliasing at upper range frequencies which can compromise the statement about not hearing the shaped filters, since it can manifest as different sound/distortion in lower range.
If anybody experienced can discuss e.g. TPDF, TPDF shaped, modified-e-weighted, low-shibata etc. it could be probably beneficial.
Best,
Jan
Noise shaping was preferred with the poor quality 16 bit converters of the past, but with modern converters it's not really necessary. A lot of professional audio engineers now stick with TPDF, for the aforementioned reason, as well as for the fact that if any further processing is done to a lossless file (including encoding to a lossy format, or even a gain change), the effects of noise shaped dither will be completely irrelevant and could lead to audible distortion. With no shaping/TPDF it's less of a problem.
If you really wanna get your head around dither, then I can highly recommend Ken Pohlmann's book, "The Principles of Digital Audio". If you read, absorb and understand that, you'll be in a much better position to discuss it.
Alright, maybe I am not as advanced audiophile as the others here, so my posts could be searching for knowledge rather than proving it.
Hydrogenaudio is full of knowledge. Do search more, there is much to find.
Anyway I tried the SoX command line program and find its dither options promising.
Also it is discussed (http://www.audiocheck.net/audiotests_aliasing.php) that some cheaper (but common) sound card can exhibit playback aliasing at upper range frequencies which can compromise the statement about not hearing the shaped filters, since it can manifest as different sound/distortion in lower range.
Now that you became an expert please explain how aliasing makes shaped dither more audible? Samples welcome.
If anybody experienced can discuss e.g. TPDF, TPDF shaped, modified-e-weighted, low-shibata etc. it could be probably beneficial.
Been there, done that. Even posted about my findings and i use low-shibata, others may like others more.
Please tell us how all the above shapes sound to you, offer samples and where it is problematic and audible, come back when you've done that and surprise us.
I really like to have a harder #TOS8 enforcement because many noobs palaver around to much.
Noise shaping was preferred with the poor quality 16 bit converters of the past, but with modern converters it's not really necessary.
Modern converters are better than 20bit these day easily it seems. So in that case no one shapes.
In the case here we talk about foobar using shaped dither when going to 16bit from HiBit content.
We can leave out if lossy encoding wastes bandwith for representing the dither.
I'd like to see a sample of any commercially available title that demonstrates an audible undoing of noise-shaped dither with lossy encoding, let alone a simple change in gain.