Extraordinary claims etc. The web site is full of nonsense.
However, if Bybee were happy to send some samples out for test, fine. But all things being equal, there's no point in buying what is most probably snake oil, it's incumbent on Bybee to facilitate proving their claim. But as far as I know they are not prepared to do this. No surprise there...
Isn't it a 16bt file delivered on a 24bit system ?, how is that different to a 16bit file delivered on a 16bit system ?.
Dan.
A 16bit LSB is 8 - 24bit LSB's, I'm not about to figure out what all went into the first file you posted but it was not properly dithered because there are long runs with no change, this is the artifact of undithered music (it suddenly fades to virtually nothing with high distortion). I have to turn up the volume during very quiet passages to hear it, I will allow for the fact that some very experienced listeners might not have to.
Maybe some Audacity action on the file? This is typical, trivial errors make ground to strange hypothesis.
John, I really hope you catch up and enjoy some time together, I sure would like to have a lunch time chat with both of you, please give him my encouragements.
Dan.
So, how do I convert 16bit files to 24bit ?.
Dan.
YES, 'trivial errors' like DC across transformers make strange observations and questions and peer review got you the answer. I want to make a loopback recording at 24bit res and starting with a 16bit res file...how would you do it ?.
Dan.
What you did is fine. Dither was grayed out because any dither that was present in the original recording is still preserved. Merely changing the file to 24b-bits only adds some zero padding at the LSB end of the words.
If you had modified the 16-bit file in Reaper in some way that would make dither appropriate, then the option would not have been grayed out.
If the resulting 24-bit file is undithered at 16-bits, it means the original 16-bit file was undithered to start with.
Yeah, what I thought, that's why I'm asking but I am keen for correction. So, if my 16bit>24bit conversion method is correct then the loopback recordings are valid too, which means the subjective differences are valid too. As pictured in my post with both tracks loaded into Reaper it is simple to instantly switch between the tracks on the fly and compare and in this environment I find the differences are clearly apparent. Dan.
I have tried coatings and found them to be good, IMO on aluminium might be a mistake though.
Dan.
Dan.
Scott really did waste his time analysing those files, what a surprise. Cleared it up for me though, thanks Scott
Last edited:
But the second file looks to be properly dithered but upsampled to 24bits it in fact has a 6dB higher noise floor (exactly BTW). Please don't speculate as to what is going on, the software has no way of knowing if any pre-processing was done properly.
Dithering after truncation is not useful. If truncation is to be performed by Reaper then by default it dithers unless someone unchecks the box so as to disable it. Thus, not clear to me what preprocessing we might be talking about. Preprocessing prior to importation into Reaper? Or, preprocessing by plugins performed in Reaper that resulted in extended bit-depth below 16-bits? I ask because I have plenty of experience with Reaper and dithering, and I am trying to make sense of how Reaper might gray out the dither option if dithering should be applied.
Last edited:
I noticed the resistance of the BYEBEE core is the same as the skin-effect equiv R. Of course skin-effect is 90 degrees out of phase with LF --- in a transformer coupled or balanced symetrical circuit there could be some noise reduction at highest audio freqs when phase shifts cancel. The BYB R thingy may only need to be nearly the same R as skin-effect R.
THx-RNMarsh
Do I understand this correctly... you should dither only at the end or last/final step?
-RNM
THx-RNMarsh
Do I understand this correctly... you should dither only at the end or last/final step?
-RNM
Well, with certain signals the hash goes down to below 10kHz easily. And it's tones, not noise.
I would think only the general baseline noise increase we see after 20kHz is that noise shaping, but those discrete frequencies are something else, with no explanation coming to my mind. I need to try measure some more DACs to see if this is a more general theme with Delta-Sigma-DACs...
Dithering should only be performed when needed. It is needed immediately prior to bit-depth reduction. Thus, if one wanted to mix down audio data with bit depth greater than 16-bits to CD, just before truncation removes lower bits below the 16th, dithering should be performed. By that means some lower level bit information is preserved in the resulting 16-bit data, but below the noise floor. If dither were to be applied after truncation, the lower level bit information would unavailable for preservation under the noise floor.
Last edited:
I would think AKM have used sync'd unwindowed FFT with the AP being the source and analyser. We cannot know for sure, though.
As for the skirt around the base of a spectral line, in synced loopback, I always find those even at way lower levels than 0dBFS once enough averaging is applied, be it magnitude-only of a large size FFT's output or with time-domain averaging before FFT, sometimes both to really dig in deep (alas, takes hours sometimes).
With the RME, ADC and DAC chips run off the same clock lines and convert at almost the same point in time so this skirt should not be any sort of jitter, but then, what is it?
- Status
- Not open for further replies.