What ? Eh ? Run that by me again ?

I remember when Technics started on their '1 bit' Mash version - which clearly wasn't one bit, but was over sampling. They tried to explain it at numerous staff events but never really managed a grown up explanation.

I've you want to see how daft it all was - this link tries to help but totally confuses. We sold loads of these by totally confusing the customers. Technics said it was better, supplied loads of technobabble and we told the customers it was better. In truth, I couldn;t hear any difference at all with these new Mash labelled machines and not once did I get told why it was better in language I could understand. Technics failed to explain it. They tried their sales division who clearly tried but were just as confused as we were and question and answer sessions got a bit out of hand with the specialists effectively disagreeing, so they brought out some learned, well qualified technical bods who talked a totally different language, and now, with hindsight, didn't understand it either.

What do you think of that article from 1990?
 
Slouching Raymond said:
That assumes that you are taking perfect analogue samples. But we're not, we're digitizing them into steps.
Click to expand...
Bit unfair and incomplete Ray? I DID say that the THEORY was correct but hardware is never perfect. We do not "take slices" the A/D process is a series of discrete voltages.

Nothing on the 44.1kHz v 48kHz response limits yet?

Dave.
 
ecc83 said:
Do we need to dither 24 bit recordings? The theoretical dynamic range is 144dB. Not that good in practice of course but still better than THE very best converters which only get a DR of the low 120dBs. I always understood that 24 bits 'dithered' on the unavoidable analogue noise floor.
Click to expand...

The theoretical dynamic range is 144 dB, but it can only exist in digital. For everything else (the analog realm of mics amps and speakers) there is the thermal limit of around 120 dB, as you know. Physics and all that.

Do we need to dither 24 bit? Is "self dither" a thing beyond the thermal limit?

Yes and no respectively. Dither creates noise and erases or removes a destructive form of harmonic distortion near the zero crossing of the waveform. That means very low signal levels approaching the least significant bit (LSB).

Why is it necessary? And why is it necessary to bias a tape?

Linear Transfer Function.

Dither means decoupling the LSB from the signal. TPDF dither at 2 bits wide is like controlling the LSB by rolling a pair of dice. It creates noise, eliminates the distortion effects of quantization error and enables linear transfer. Simply adding noise instead does not do the other 2 things. If the LSB is not random, it's not dither. Claude Shannon got it right.
 
Things have gone astray a bit . . . I've got one that makes me laugh every time I read it--some mixer/producer says that they're going to add some "grit" to a recording they're working on. What on Earth does "grit" mean here? I think I know, but maybe not. Maybe it means distortion. Over at SOS virtually every mixer says this about a song they mixed, usually at the end while they're describing the many, many plugins that are on the main bus or other busses or tracks. "The Behemoth Maximizer is in there doing very little, just adding some grit before it hits the Baffling Descrambler, again doing very little, just pushing the track to the very edge . . . ." For the record, I don't think I've ever messed around with either polarity or phase really ever, and I always record at 48kHz because when I first started out that was the next "best" rate that the machine could handle, above the old 44.1. When I learned that video is typically shot at 48 kHz, I felt vindicated.
 
Last edited:
TimOD said:
Things have gone astray a bit . . . I've got one that makes me laugh every time I read it--some mixer/producer says that they're going to add some "grit" to a recording they're working on. What on Earth does "grit" mean here? I think I know, but maybe not. Maybe it means distortion. Over at SOS virtually every mixer says this about a song they mixed, usually at the end while they're describing the many, many plugins that are on the main bus or other busses or tracks. "The Behemoth Maximizer is in there doing very little, just adding some grit before it hits the Baffling Descrambler, again doing very little, just pushing the track to the very edge . . . ." For the record, I don't think I've ever messed around with either polarity or phase really ever, and I always record at 48kHz because when I first started out that was the next "best" rate that the machine could handle, above the old 44.1. When I learned that video is typically shot at 48 kHz, I felt vindicated.
Click to expand...
Clipping, squaring off the tops a bit I think. Distortion and saturation add harmonic content as well as squaring wave forms. Clipping just squares stuff. I would call that grit. In most of the ads for software "grit" is often used to describe "preamp saturation" emulation in "channel strip" plugins.

In re dithering: the only reason I mentioned it is out of the old habit of thinking of Red Book mastering for CD 44.1 16 bit. Completely unnecessary for uploading wave files to whichever streaming service is going to run your music through their codec.

I tend to upload 320 mp3 since I am too cheap to spend money on more online storage in soundcloud, that way I can post more. Part of converting to MP3 is down sampling so I do use dither for that. Don't really know if it makes a difference but Ozone says to apply dither, so I do.

As for 48 versus 44.1 , Blue Ray standard is 48 and as mentioned, it is the standard for video sound but honestly, I use just in case compatability is an issue in the future. Also IIRC since 44.1 isn't a multiple of 8 that it is more susceptible to rounding errors?

I started recording digital in ~1999 and was lucky enough to be able to record 24 bit in Pro Tools. But coming from tape my first recordings were inferior to my ears as I started at 16 bit to save disc space because I had not bought a dedicated SCSI recording drive yet. When I tried 24 bit//44.1 IMO the sound quality was much higher. I experimented by transferring tape sessions into the computer and I found I could blind test and hear the difference with different bit rates.
 
rob aylestone said:
I remember when Technics started on their '1 bit' Mash version - which clearly wasn't one bit, but was over sampling. They tried to explain it at numerous staff events but never really managed a grown up explanation.
Click to expand...
I took a look at the article, Rob.
I am familiar with Delta-Sigma conversion. I worked with it in the 70s.
A regular 16-bit converter allows for each sample to have any level within the range. It is highly unlikely though that the next sample would be at the opposite extreme to the current one.
The Delta-Sigma converter tracks the signal, and decides whether the next sample is either above or below the current sample, and generates a '1' if above, or a '0' if below.
To convert back to analogue, you simply feed the digital stream of 1s and 0s through an resistor-capacitor integrator.
Both the encoder and decoder have a resistor-capacitor integrator.
The output of the integrator looks like a sawtooth, and continually fluctuates above and below the original signal, but generally tracks the signal. There is always an error.
The Delta-Sigma converter must operate at a much higher sampling rate than a normal pcm converter, to achieve the same quality of signal.
When a higher sampling rate is used, the sawtooth signal which the integrator generates, has much finer teeth, and therefore tracks the input signal more closely.
 
Slouching Raymond said:
Nonsense, that's just how it is.
Click to expand...
Not at all and I stand by the fact that you left out my qualifying statement. Has anyone found a converter specification yet that gives two filter frequencies? One for 44.1 and a higher one for 48kHz.

I may just be an old valve amp jockey but I won't be bullied.

Dave.
 
Papanate said:
Are you talking about advanced filtering?
Click to expand...
I am simply saying that the only advantage that 48k sampling appears to have over 44.1k is that it has a higher theoretical upper audio frequency response. However, does that obtain in practice or, since few people can hear 20kHz leave alone anything higher AND few microphones even reach 20k, I suspect the response in converters is simply made for 44.1kHz use and stays at the same corner frequency for 48kHz. Unless someone knows better?

Dave.
 
ecc83 said:
I am simply saying that the only advantage that 48k sampling appears to have over 44.1k is that it has a higher theoretical upper audio frequency response.
Click to expand...
It's not theoretical - 48 kHz does have and extended audio range.


ecc83 said:
However, does that obtain in practice or, since few people can hear 20kHz leave alone anything higher AND few microphones even reach 20k, I suspect the response in converters is simply made for 44.1kHz use and stays at the same corner frequency for 48kHz. Unless someone knows better?
Click to expand...
The difference people are hearing, is not the high frequency content, but the fact that lower sample rates cause the converters to distort the analog signal - 48 kHz included - but it's still a little bit less than 44.1 kHz This is due to ripples in the bandpass filter cased by restricted high pass bandwidth in lower sample rates.

It’s not that these higher rates actually contain extra musical information, the issue is to do with the filters playback systems need to use to decode digital. Higher rates allow playback systems more room to work, and many will sound better as a result. That's a reason to record at 96 kHz 24 Bit.
 
Papanate said:
It's not theoretical - 48 kHz does have and extended audio range.



The difference people are hearing, is not the high frequency content, but the fact that lower sample rates cause the converters to distort the analog signal - 48 kHz included - but it's still a little bit less than 44.1 kHz This is due to ripples in the bandpass filter cased by restricted high pass bandwidth in lower sample rates.

It’s not that these higher rates actually contain extra musical information, the issue is to do with the filters playback systems need to use to decode digital. Higher rates allow playback systems more room to work, and many will sound better as a result. That's a reason to record at 96 kHz 24 Bit.
Click to expand...
I can understand the much higher sample rates can be better (like higher tape speeds) but 96kHz is a massive jump from 44.1kHz and I assume people genuinely can hear an improvement to justify the bigger file sizes. My point is whether 48k is really audibly, provably better?

I did not say 48kHz was just THEORETICALLY extended. Of course it has the capacity to be, a bit. What no one has yet demonstrated (a mncftrs spec will do) is that any particular interface actually gives that bit extra HF when running at 48k or do they just use a fixed F filter for both? I strongly suspect the latter.

Dave.
 
ecc83 said:
I can understand the much higher sample rates can be better (like higher tape speeds) but 96kHz is a massive jump from 44.1kHz and I assume people genuinely can hear an improvement to justify the bigger file sizes. My point is whether 48k is really audibly, provably better?

I did not say 48kHz was just THEORETICALLY extended. Of course it has the capacity to be, a bit. What no one has yet demonstrated (a mncftrs spec will do) is that any particular interface actually gives that bit extra HF when running at 48k or do they just use a fixed F filter for both? I strongly suspect the latter.
Click to expand...
48 kHz doesn't give you more frequencies - 96 kHz doesn't give you more frequencies - 96 kHz gives you no distortion that 44.1 kHz does - you hear anything it's cleaner audio - at 44.1 kHz you already exceeded hearing levels - 48 kHz is little bit better on the distortion levels - but 96 kHz and above are zero distortion. What is a F Filter?
 
Papanate said:
48 kHz doesn't give you more frequencies - 96 kHz doesn't give you more frequencies - 96 kHz gives you no distortion that 44.1 kHz does - you hear anything it's cleaner audio - at 44.1 kHz you already exceeded hearing levels - 48 kHz is little bit better on the distortion levels - but 96 kHz and above are zero distortion. What is a F Filter?
Click to expand...
Frequency.
 
This Monty Montgomery video is pretty good about the digital vs analog "difference" and shows that an analog waveform is reproduced perfectly from a digital source. That doesn't mean that due to other various reasons from the manufacturing process to the playback device and everything in between, that a digital copy (like a CD) may sound different than an analog copy (like a record) anyway.
 
You must log in or register to reply here.
Back
Top