How is it possible for me to use 26 letters to reproduce 44 phonemes?
They are discrete characters in theory, but as anyone trying scanning can relate there is close to infinite variations, yet they are used to reproduce a semi time discrete tonal combination?
Which is more important, discrete time or discrete level?
They are discrete characters in theory, but as anyone trying scanning can relate there is close to infinite variations, yet they are used to reproduce a semi time discrete tonal combination?
Which is more important, discrete time or discrete level?
Last edited:
Badly is the usual answer
Bough/Cough
Cow/Bow
I live in Reading but if I am reading a book I pronounce it differently
And if a boat of rowers have a row over which way to row!
And Romanising Hindi words is a nightmare all of its own!
The fact that English works in any form as communication is amazing!
Bough/Cough
Cow/Bow
I live in Reading but if I am reading a book I pronounce it differently
And if a boat of rowers have a row over which way to row!
And Romanising Hindi words is a nightmare all of its own!
The fact that English works in any form as communication is amazing!
It doesn't seem to be a problem with bit-depth. It is jitter related in the following way: if I feed ultra low jitter digital audio into the dac, I can set ASRC DPLL bandwidth arbitrarily, because it remains stable in all cases. What I find is that with minimum DPLL bandwidth the reverb tails are mostly there, but if I increase DPLL bandwidth the least amount I can, I start really losing the tails. What happens when DPLL bandwidth is low is that the VCO creates the least of its own jitter, and as bandwidth increases so does the internally created jitter.
Not clear that bits are being trucated exactly. Bit-depth related resolution of other sounds is still very good. I am still thinking about what could be going on. In a very speculative way I wonder if something in the S-D modulator is losing them because they are more stationary than other musical sounds. Maybe it sees them as something like drift offsets in the audio spectrum. To make that more clear, imagine a spectrogram view of music with reverb tails, where the tails can be visualized. They remain relatively stationary in the spectrum view with time according to the nature of how they sound, so knowing that maybe one could HP filter the spectrum to remove them as viewed in spectral domain. Just to be clear, I really am doing a lot of speculating at this point and probably need to do some reading to see if any of this could hold water.
Anyway, continuing with the story for now, it could be that jitter creates artifacts that causes the modulator to HP filter in the spectral domain. It does't necessarily have to be trying to go after the reverb tails directly.
Anyway, continuing with the story for now, it could be that jitter creates artifacts that causes the modulator to HP filter in the spectral domain. It does't necessarily have to be trying to go after the reverb tails directly.
Last edited:
Which in any recorded 24 bit depth recording you have at best approximately 3 bits of noise. (under the glaring assumption that the levels are set to use most of the dynamic range) You may well have greater noise floor from Brownian motion of gas molecules impinging on the microphone.
This whole thing, in general, isn't adding up to what I remember learning along the way. As I said though this isn't my domain of expertise, so I fully reserve the right to be wrong.
Bill, unless there's something hideously wrong, the filter curves I've seen from ti/akm/cirrus(Wolfson) all are benign, and would leave me doubting anyone could hear the difference. Exception being perhaps different folks's slow roll off implementation on 44.1 or 48 where both hf rolloff and aliasing are potential concerns. There's a simple solution to that one, though: don't use those filters!
SNAP! (well, to #1&2) the oxidization inherent to all contact systems especially those with nickel needs to be burnished and the "preservative" oil in DeOxit seals oxygen from the burnished contact pair. I also agree with the DeOxit for #3. Can you do more than tease us regarding the end-to-end suggestion?
Cheers,
Howie
Howie: I have my bottle of deoxit, but every full moon is too much. It did fix up a tv remote once where one of the buttons had stopped working.
Daniel: I was meaning more the time domain response of the digital reconstruction filter. Take for example Chord Electronics Qutest D/A processor Measurements | Stereophile.com where chord offer you 'Incisive Neutral' or 'Warm' Filters with different time domain responses. In their case roll off is extremely sharp so no aliasing. But they will not produce exactly the same output. I see this as as likely to be audible has hash 120dB down.
Daniel: I was meaning more the time domain response of the digital reconstruction filter. Take for example Chord Electronics Qutest D/A processor Measurements | Stereophile.com where chord offer you 'Incisive Neutral' or 'Warm' Filters with different time domain responses. In their case roll off is extremely sharp so no aliasing. But they will not produce exactly the same output. I see this as as likely to be audible has hash 120dB down.
Okay, yes we're talking the same thing in different mindsets. And I agree with the assessment, but that will not stop people from casually mentioning a huge difference.
Hey Bill,
I agree; I thought Ed said once per blue moon, once again I am guilty of not reading carefully. I typically burnish/preserver patch cables once per year or two which works well for me. If one lives in a humid and corrosive atmosphere like a seaside home, more frequent may be warranted.
Howie
- Status
- Not open for further replies.