You figured it out fast, serious vinyl archiving can be a very tedious process/ritual.If the card connects to SoX as an I/O device you are done, as you say for home use the latency is a don't care. You can record from SoX and pipe the result to the RIAA plus any other filter and pipe that to the play feature.
Realistically, 20 bits is the absolute maximum available with PCM digital.
It would be nice if the Universe would stop all that 'Brownian' noise and allow, say, true 24 bit or greater, but .......
The problem is that distortion in PCM digital increases as level decreases (the opposite of analogue).
I'm so sick of hearing people assert that 16-bit PCM has a "90 dB dynamic range" or even "98 dB with dithering", it doesn't - those are S/N numbers. By -80dB distortion in 16-bits is in to WHOLE DIGITS, and this where the most the fragile part of a recording (music) often lies (and will lie in our theoretical digital RIAA engine if the preamp isn't turned up high enough) - the subtle harmonics, 'decay' and 'ambience cues'.
And don't even get me started on sample-rates, digital/anti-alias filtering .....
Indeed. Properly dithered 16-bit digital has a dynamic range of almost 120 dB - way beyond any vinyl.
Please explain that claim.
I don't believe anyone is proposing setting their soundcard ADC to 16 bits for digital RIAA
and analog channels are measured in THD+Noise too - noise is the practical limit
the best delta-sigma converters have differential linearity below their electrical noise floor when looked at with long averaging/narrow bandwidths irrelevant to human hearing's 100 Hz or larger critical bands (~300 Hz at our 3kHz peak hearing sensitivity) to verify this linearity - but in practice it is pointless for perception when the noise in each critical band is what matters
and analog channels are measured in THD+Noise too - noise is the practical limit
the best delta-sigma converters have differential linearity below their electrical noise floor when looked at with long averaging/narrow bandwidths irrelevant to human hearing's 100 Hz or larger critical bands (~300 Hz at our 3kHz peak hearing sensitivity) to verify this linearity - but in practice it is pointless for perception when the noise in each critical band is what matters
Last edited:
No - but my point is that as even 16 bits is more than enough to cope with the dynamic range of recordings on vinyl (actually 12 bits would probably be enough), we have quite a lot of headroom when using 24 bits, and thus don't really have to worry about using every single dB of gain optimally.
Equalized, yes, but there's ~40dB or so for EQ that must also be taken into account.
Yes but the self dither is at the source, at the low end -20db signal is still well dithered by the surface noise (currently my opinion but I will try to prove it). At the high end you win and get the benefit of pre-emphasis. I claim you never see the quantization noise. The Pure Vinyl folks wrote a nice article but they dealt with the noiseless case.
Last edited:
Good work, though there are some dependencies. I still use XP and ASIO drivers since I'm too cheap to buy newer devices.
At -78dB 16-bit PCM is down to 3 bits. I don't have the maths skills to work out what distortion would actually be in a 3-bit PCM system, but I'll guess it will be into double digits.
I'm aware that the use of dither and over-sampling ameriorates this.
I'm curious as to how 16-bits can somehow become 20-bits/120dB (S/N - if a reasonable limit is set for allowable distortion in calculating 'dynamic range' then that demonstrated by PCM audio at the bottom 3-4 bits should not be included in the number).
Are you referring to HDCD? That is, in effect, a kind of 'compander' system (analogous in some ways to some Dolby systems), not true 20-bit.
And the question begged is; why did the developers see a need to 'improve' on 16-bit PCM in the first place?
ETA - dynamic range of vinyl. Vinyl at best has a S/N of <70dB. However, the noise is mostly below a couple of KHz and it can reproduce audible content in the mid to HF region well below its noise-floor, at far lower distortion than 16-bit PCM does at such low levels, and a suitably 'weighted' S/N measurement results in a dynamic range of at least 90dB.
Last edited:
Good work, though there are some dependencies. I still use XP and ASIO drivers since I'm too cheap to buy newer devices.
... but, Windows 10 Preview and MS Virtual PC (or Oracle's VB) are free ...
You're confusing distortion and noise, unless you're assuming that there's no dither from the phono, which is a bad assumption. At -78dB, you're already well into cartridge thermal noise, even if it's not down on a record.
No. I guess part of the problem is that you are confusing distortion with quantisation noise. The maths part is important - people who rely on "intuitive" understanding of sampling theory and digital systems usually get it badly wrong.
No, I am not talking about HDCD or any sort of compander. I am talking about 16-bit linear PCM, where the SNR over the whole frequency range will be 96 dB, but individual tones (if dithered correctly) will still be audible at levels below 110 dB.
Here is a great illustration: xiphmont: The dynamic range of 16 bits
To have recording and processing headroom.
ESS whitepaper has -60 dB sine fft plot showing spot noise floor ~100 dB from the -60, any sum of distortion bins had to be less than -90 dB below the -60dB sine since there are no visible harmonic peaks in the plot
John Atkinson's Stereophile measurements of 6-7 year old DACs in universal players show -90 dB sine with just noise in several reviews
John Atkinson's Stereophile measurements of 6-7 year old DACs in universal players show -90 dB sine with just noise in several reviews
Last edited:
As I'm not an engineer I tend to use terminology incorrectly sometimes.
But I do understand that strictly speaking the only 'noise' in PCM capture is distortion produced by quantisation and steep digital filtering. Nonetheless I've always felt that rather disingenuous use has been made of "the numbers" for PCM (specifically, 16/44), starting with the marketing hype at its introduction into the consumer market.
I'm actually not dissatisfied with 16 bits for a 'delivery format', I feel that sample rates are more important - I sometimes convert 24/88.2 and 24/96 recordings to 16 bits and archive them them that way.
The simple fact is that ultimately everything - all 'legacy' audio-visual media - will have to archived digitally or it will eventually be lost. In fact many experts are concerned about a looming 'legacy format' crisis, decaying media and decrepit, unserviceable hardware. Ironically, the one possible exception is vinyl records, for which this crisis will much further in the future.
I think it's a great shame the introduction of a consumer digital format wasn't delayed for a few years until it was more mature, and a higher sample rate used, but I guess that would always have been a pipe-dream.
Last edited:
But it isn't distortion, it is noise. It sounds just like tape noise.
Did you read the link I posted?
In what way?
Why? Can you hear beyond 20 kHz? Do you think there is significant information beyond 20 kHz on a vinyl record?
And in what way do you feel it wasn't mature enough? What benefit do you think a higher sample rate would bring?
its amazing when you can point to hard evidence - distortion measurements deep under the noise floor by multiple tests, from datasheets, prosumer soundcard plots in multiple forums to independent reviews - and the concepts aren't grasped and the naïve misunderstandings just repeated louder
we all start from a position of ignorance - but some have a willingness to learn
we all start from a position of ignorance - but some have a willingness to learn
Last edited:
Funny thing is, 2-bit PCM can capture 'a tone' at any frequency under the nyquist limit (although only at 0dB, I guess)
For example, a 12KHz sine at 02/48;
0000111100001111100001111100001111 ........etc' (yes, I know this oversimplifies, MSB used for phase etc')
Oh, it's not a sine when it comes out of the ADC/DAC process?
Just over-sample 16x!
Now let's try recording some music with this 02/48 PCM .......
ETA >> An epiphany! With dither, a 'tone' at -6dB could be captured and reproduced!
For example, a 12KHz sine at 02/48;
0000111100001111100001111100001111 ........etc' (yes, I know this oversimplifies, MSB used for phase etc')
Oh, it's not a sine when it comes out of the ADC/DAC process?
Just over-sample 16x!
Now let's try recording some music with this 02/48 PCM .......
ETA >> An epiphany! With dither, a 'tone' at -6dB could be captured and reproduced!
Last edited: