In what way should an latency of 100ms. (so long as it's constant delay versus frequency) be considered an imperfection for the purpose of home playback? Before you answer, keep in mind that an CD (or an vinyl album for that matter) could easily have an playback latency measured in tens of years with respect to the original recorded event. Causality demands that any recorded event must exhibit latency upon playback, so I'm not sure where you've come by the notion that this counts as an in perfection. Editing, and lip synching for video requirements are a different matter than for audio only home playback
Last edited:
I think your idea of perfection requires infinite bandwidth. In the real world that is not possible, as it would require infinite power to run the circuits. Given the more realistic aim of reproduction of a band-limited signal then sufficiently fast sampling with a good reconstruction filter can achieve it.Kastor L said:
Why do people often demand of digital audio things they would never demand of analogue audio: infinite bandwidth, zero distortion, zero noise (so needing cryogenic circuitry) and acausality (so requiring a time machine)? It is obviously daft to require these things of analogue; it happens to be equally daft (but perhaps less obvious) to require these things of digital.
If only the reccording devices & process could have the same paranoia level of "perfection" we are looking for here !
We are limited by the equipments than each Studio has like the talent of the sound "engineer" for the reccording capture & mix for making a disc ! Everybody is not Chesky !
We spend a lot of time to make sounding better the limitations before without changing this lasts on the reccording side !
Simple thought... maybe false !
We are limited by the equipments than each Studio has like the talent of the sound "engineer" for the reccording capture & mix for making a disc ! Everybody is not Chesky !
We spend a lot of time to make sounding better the limitations before without changing this lasts on the reccording side !
Simple thought... maybe false !
Setting aside the level of transparency of the microphone and the ADC, I was referring to the IR after the ADC, if we can zoom in on it and see all of it precisely.
Here is some text on microphones and ADC I saw today — The Fifth Element #78 Page 2 | Stereophile.com
Do you have thoughts on how an impulse response looks in analog cassette tape? Your question seems a bit loaded I'm not sure.
______
No, that doesn't seem true at all. Perhaps you need to reclarify.
See text here — dsd
This as well — http://www.cirrus.com/en/pubs/whitePaper/DS668WP1.pdf
Plus this — http://www.audio-trek.net/diyAudio/MinPhaseIntro/MinimumPhase.htm
______
I'm concerned about pre-ringing in all sines, not only in perfect square-waves or in ultrasonic sines.
This isn't an ultrasonic discussion the way you paint it.
Reality is always minimum-phase, unless you're recording a recording or you're recording an unusual speaker which has linear-phase events in it or similar unusual events like that.
You seem to be confused, really.
A maximum-phase filter, is always pre-ringing with no post-ringing.
What does the pre-ringing, in that case, have in connection to Nyquist or ultrasonics?
In that case its the impulse of the AAF you'll be seeing. Most ADCs these days are S-D architecture meaning the effective AAF (due to high oversampling ratios) is a linear phase FIR filter.
I was referring to R2R tape, not cassette. No idea, no.
I don't know what AAF is, I'll look that up, thank you.
Cassette tape sounds pretty good to me, pretty happy with that sound.
Cassette tape sounds pretty good to me, pretty happy with that sound.
Anti-aliasing filter - used to ensure nothing beyond fs/2 gets through to the sampling circuit.
When I think of latency, I usually think of connecting a Midi synthesizer or an electric guitar to a computer.
If you hit a key on the keyboard, you want it to make a sound when you hit the key, not 100ms later.
That's why ASIO or KS is paramount in some areas, but it certainly can't fix the error in a reconstruction filter.
ASIO is needed in some PC games as well.
After that, we have audio/visual, like watching fast action sports, you want the physical action to connect with the sound, then it looks much more real.
For "audio only" playback I don't suspect it causes very much imperfection.
The only area I can think of would be the hardware or software code, or the hardware components, causing the imperfection in that case.
For example quite a few people say media players, i.e. software code and ASIO or KS sound different.
In the case of KS that would be due to the "direct stream" of the data without buffering. It becomes a "straight river" of data instead of a "complicated aqueduct", but this is the area when audio starts to become far-fetched in how electron transmission affects the sound-quality.
You mean digital is more perfect than reality, or more perfect than vinyl and tape?
That is why DSD usually sounds better than CD's, the studios which use DSD are "paranoid" and use better microphones and ADC as well.
Remember that "garagerock" received it's name from being recorded in a garage.....
Most likely with not very good recording equipment......
Yet it still sounds pretty good imho, LoL.
In the extreme, perfect sense of perfection, then yes.
I only have a basic Sony pocket recorder, I don't record, but if I did then I would consider DSD or DSD-Wide with an Audio-Technica AT-8015 microphone.
With playback I'm inclined to believe we need video-like transient response to achieve a more surreal sound.
For example one of the best sounding DAC's I have ever heard was using an AD828 video chip.
Claiming that the thickening of spectral lines from DSD IM is an advantage seems like making a virtue of necessity. It also seems to be confusing the basic issues of a technique with the typical implementation problems of a technique.
Claiming that misuse of opamps in DAC circuitry (e.g. as I/V converters, or active filters) is somehow a fundamental flaw in DACs is missing the point. Passive pre-filtering can slow the signal seen by the opamp down to the region where the opamp is well behaved.
This thread now appears to be casting around for a new purpose, given that the original question was meaningless. It is fast heading for the usual Fourier/Shannon/Nyquist/causality denial paradigm.
Claiming that misuse of opamps in DAC circuitry (e.g. as I/V converters, or active filters) is somehow a fundamental flaw in DACs is missing the point. Passive pre-filtering can slow the signal seen by the opamp down to the region where the opamp is well behaved.
This thread now appears to be casting around for a new purpose, given that the original question was meaningless. It is fast heading for the usual Fourier/Shannon/Nyquist/causality denial paradigm.
The original question was pertaining to amplitude resolution and time resolution.
Without any kind of noise floor at all then the bit depth is either infinite or it varies from silence to the loudest sound possible.
However the dynamic range of the music itself will only vary 96 dB with 16-bit / 44.1 kHz music.
With oversampling you can achieve the equivalent of 17-bit resolution or 102 dB.
However that is only the dynamic range to the quantization noise floor, it's still 96 dB even though it's stated as 102 dB.
Apparently with shaped dither you can achieve 120 dB dynamic range with 16-bit audio, however it's still 96 dB at the same time.
There are multiple "dynamic ranges" for a single bit-depth, at the same time, sources like Wikipedia and everywhere else I checked make zero distinction for this.
We can also avoid quantization error and dither entirely by simply making a 16-bit file via software alone, like in Reason, what is the dynamic range then?
No distinction for that either.
You can call the question whatever you want, it seems more like you're utilitarian and quick to dismiss all non-textbook discussion.
Without any kind of noise floor at all then the bit depth is either infinite or it varies from silence to the loudest sound possible.
However the dynamic range of the music itself will only vary 96 dB with 16-bit / 44.1 kHz music.
With oversampling you can achieve the equivalent of 17-bit resolution or 102 dB.
However that is only the dynamic range to the quantization noise floor, it's still 96 dB even though it's stated as 102 dB.
Apparently with shaped dither you can achieve 120 dB dynamic range with 16-bit audio, however it's still 96 dB at the same time.
There are multiple "dynamic ranges" for a single bit-depth, at the same time, sources like Wikipedia and everywhere else I checked make zero distinction for this.
We can also avoid quantization error and dither entirely by simply making a 16-bit file via software alone, like in Reason, what is the dynamic range then?
No distinction for that either.
You can call the question whatever you want, it seems more like you're utilitarian and quick to dismiss all non-textbook discussion.
Ah that's not engineering dynamic range, that's perceptual dynamic range. Easy to confuse the two but they're not equivalent - with shaped dither the engineering (measured) dynamic range will actually be lower than 96dB by quite a bit seeing as the shaping can only move noise around (within the 22kHz bandwidth) it can't remove it. The shaped dither gets its increased perceived dynamic range by moving noise away from the 2-4kHz band where the ear is most sensitive and piling it up at higher freqs (where the ear isn't so sensitive).
- Status
- Not open for further replies.