Well I just don't think you would have the flexibility of DSP in the analog world. Any adjustment you make pretty much = more uncorrelated noise. With digital and photography you can do all sorts of gradient mixing between the 3 photos to take out the clipped portions and the undersaturated. How do you plan to selectively take out the bad stuff while leaving the good with analog audio? And how could this improve upon conventional mixing? Conventional mixing already will use dynamic range tricks like parallel compression with low pass filtering to fill in the low end while leaving dynamics in places.
I have tried to think of some ways in which HDR parallels with audio - I have seen some people use cluster micing but I am not so sure this is the same either. It just seems like HDR works best with things that don't change in the time domain - unlike sound.
I have tried to think of some ways in which HDR parallels with audio - I have seen some people use cluster micing but I am not so sure this is the same either. It just seems like HDR works best with things that don't change in the time domain - unlike sound.
Last edited:
Entirely agree. But that's not even a goal here. The goal of the HiDRA concept is to maximize the recording and playback fidelity of analog tape. Nothing more.
We're not really adjusting anything here, just selectively cutting back and forth on the fly to stay in the S/N sweet spot, so to speak.
Please re-read posts 1, 2, 13, 14.
Not entirely sure what you mean here, but as I mentioned previously, I don't think other forms of noise reduction would be out of place on a HiDRA machine.
Huh?
Last edited:
Unless you somehow find a magic sweet spot where the sounds will line up perfectly you are going to introduce distortions. And isn't the point in the first place that the three captured events will be slightly different? Chances of the syncing up are practically none as far as I can tell and you propose no solution on the mic end to capturing a sound 3 times with 3 different dynamic range sensitivities perfectly in time/phase.
The closest scenarios I can come up with to improve sound resemble a stitched panoramic photo more than HDR.
Ah! Now I see the misunderstanding.
I am not proposing multiple mics. I am proposing 1 mic/pre whose output is split multiple ways with different levels of attenuation and recorded to multiple tracks on a single piece of tape.
Thus, there are no time or phase differences between tracks, just different levels. When those level differences are compensated on playback, their in and out points will match to a small fraction of a dB.
Makes more sense now? 🙂
I am not proposing multiple mics. I am proposing 1 mic/pre whose output is split multiple ways with different levels of attenuation and recorded to multiple tracks on a single piece of tape.
Thus, there are no time or phase differences between tracks, just different levels. When those level differences are compensated on playback, their in and out points will match to a small fraction of a dB.
Makes more sense now? 🙂
Last edited:
Not really haha. I want it to though 🙂
I believe with HDR what is happening is that you have the lens set to 3 different aperture settings allowing a different amount of light in with each shot. This in the audio world is like using 3 mics. Ideally this process would take all 3 pictures simultaneously but you will find that most people are using it on still images or time lapse where they can just take 3 pictures in rapid succession.
What you are proposing really isn't any different than conventional color correction where you add synthetic contrast or in the audio world this would be mixing. I just don't see how this would cleanly work sorry.
There are some weird ways in which to capture a direct sound from electronic instruments and then reamp it at 3 different volumes with a stationary mic which might get closer to what you are aiming for but not exactly what you are expecting to happen will happen.
I believe with HDR what is happening is that you have the lens set to 3 different aperture settings allowing a different amount of light in with each shot. This in the audio world is like using 3 mics. Ideally this process would take all 3 pictures simultaneously but you will find that most people are using it on still images or time lapse where they can just take 3 pictures in rapid succession.
What you are proposing really isn't any different than conventional color correction where you add synthetic contrast or in the audio world this would be mixing. I just don't see how this would cleanly work sorry.
There are some weird ways in which to capture a direct sound from electronic instruments and then reamp it at 3 different volumes with a stationary mic which might get closer to what you are aiming for but not exactly what you are expecting to happen will happen.
Okay, feel free to call it mixing if you like. I won't be offended. 🙂
Your skepticism is noted. Thanks for volunteering your input.
Clarification on dynamic range
I just wanted to take a moment to step back and clarify something for readers of this thread.
When I say the HiDRA concept is aimed at expanding dynamic range to 100+dB, I acknowledge that this is not exactly the same as a theoretical recording system that has 100+dB of useful dynamic range per track. In such a theoretical system, every moment of a recording can contain high-quality information about both an ear-shattering clash of symbols and the quiet shuffling of audience members' feet. (Never mind that, during the symbol clash, no one will ever hear the feet shuffling.)
A HiDRA system is built upon a core hardware/software system of tape machine and tape. At any given moment, a HiDRA system has the instantaneous S/N ratio of this core system, probably limited by its tape stock to about 80dB of dynamic range.
However, HiDRA's advantage would come in its ability to dynamically drop that noise floor during the quiet passages, when subtle details are not swamped by crescendos, and a listener might actually be able to perceive them.
I just wanted to take a moment to step back and clarify something for readers of this thread.
When I say the HiDRA concept is aimed at expanding dynamic range to 100+dB, I acknowledge that this is not exactly the same as a theoretical recording system that has 100+dB of useful dynamic range per track. In such a theoretical system, every moment of a recording can contain high-quality information about both an ear-shattering clash of symbols and the quiet shuffling of audience members' feet. (Never mind that, during the symbol clash, no one will ever hear the feet shuffling.)
A HiDRA system is built upon a core hardware/software system of tape machine and tape. At any given moment, a HiDRA system has the instantaneous S/N ratio of this core system, probably limited by its tape stock to about 80dB of dynamic range.
However, HiDRA's advantage would come in its ability to dynamically drop that noise floor during the quiet passages, when subtle details are not swamped by crescendos, and a listener might actually be able to perceive them.
Bill,
What you propose is different from HDR photography:
High dynamic range imaging - Wikipedia, the free encyclopedia
Shifting by 10 dB is actually compressing 3 times, shifting by 20 dB is actually compressing 10 times the audio waveform. Imagine a full scale 100 Hz audio waveform, let's say a sinewave, refer to it as 0 dB. Then imagine that it has a -40 dB 5 kHz sinewave superposed. Now if you amplify this complex waveform by 10 dB, it will be clipped above 1/3 of the original signal voltage, and the 5 kHz sinewave will be lost where clipping occurs. So you will not gain any S/N impovement in this region. If the tape noise is let's say -50 dB, it will be just 10 dB below our 5 kHz sinal. The S/N will jump to 20 dB where you substitute the 0 dB track with the 10 dB track, but then it will jump up and down in 100 Hz rate (in our example). This solution is not practical.
Oops, now I see you probably meant the peak signal voltage as a determining factor, not the instantaneous voltage. Then, if two single clipping occurs within a short period of time, you may want to replace the whole range, not the single occurances. I think only a computerized system could make such decisions on digital signals.
What you propose is different from HDR photography:
High dynamic range imaging - Wikipedia, the free encyclopedia
Shifting by 10 dB is actually compressing 3 times, shifting by 20 dB is actually compressing 10 times the audio waveform. Imagine a full scale 100 Hz audio waveform, let's say a sinewave, refer to it as 0 dB. Then imagine that it has a -40 dB 5 kHz sinewave superposed. Now if you amplify this complex waveform by 10 dB, it will be clipped above 1/3 of the original signal voltage, and the 5 kHz sinewave will be lost where clipping occurs. So you will not gain any S/N impovement in this region. If the tape noise is let's say -50 dB, it will be just 10 dB below our 5 kHz sinal. The S/N will jump to 20 dB where you substitute the 0 dB track with the 10 dB track, but then it will jump up and down in 100 Hz rate (in our example). This solution is not practical.
Oops, now I see you probably meant the peak signal voltage as a determining factor, not the instantaneous voltage. Then, if two single clipping occurs within a short period of time, you may want to replace the whole range, not the single occurances. I think only a computerized system could make such decisions on digital signals.
Well the way I see it HDR is a form of dynamic range compression and not a form of expansion.
I see a little clearer what you are aiming for but I still think it would be loaded with distortions if you try to impliment it. But I was thinking the attentuation would come before the capture format which I now see isn't what you are saying.
So you want to use 3 or 4 tracks per every one sound you are capturing. Captured at various s/n ratios and then when playing back they would be stitched together in a composite effectively giving you a variable quality or fidelity. I could actually try this on my digital system it's just that I am pretty sure the implimentation of the crossfading between the 3 captures will be difficult and distortion prone. Also I believe clipping a signal has an effect on the non clipped portion of the signal as well as the clipped portion - in analog this can lead to drop out and I am guessing in digital it could lead to distortions.
I see a little clearer what you are aiming for but I still think it would be loaded with distortions if you try to impliment it. But I was thinking the attentuation would come before the capture format which I now see isn't what you are saying.
So you want to use 3 or 4 tracks per every one sound you are capturing. Captured at various s/n ratios and then when playing back they would be stitched together in a composite effectively giving you a variable quality or fidelity. I could actually try this on my digital system it's just that I am pretty sure the implimentation of the crossfading between the 3 captures will be difficult and distortion prone. Also I believe clipping a signal has an effect on the non clipped portion of the signal as well as the clipped portion - in analog this can lead to drop out and I am guessing in digital it could lead to distortions.
Sounds rather editing and labor extensive, as well as wasting tracks.
I think just a better machine that could utilize newer technologies would
be simpler and more user friendly. Something like a wide format EE tape,
or perhaps even wide format metal particle. Of course some heads to handle the
higher bias currents would need to be installed. Something variable bias
like the Dolby (tm) HX Pro may improve high freq. headroom as well.
I think just a better machine that could utilize newer technologies would
be simpler and more user friendly. Something like a wide format EE tape,
or perhaps even wide format metal particle. Of course some heads to handle the
higher bias currents would need to be installed. Something variable bias
like the Dolby (tm) HX Pro may improve high freq. headroom as well.
Yes, that is correct. I intentionally included an example of that in my illustration in post 14 where the first crescendo crosses into the track-3 range, comes out briefly, and then returns--but notice playback of the whole range remains on track 3. This is just to show that the switching decisions needn't be rigorously dictated by signal level, and fewer switching events can sometimes be a preferred compromise.
As cbdb kindly reminded us, we are not dealing with hard clipping events as in the digital realm. We are making switching decisions based on thresholds for the onset of tape saturation.
Valid points. My reply is that the HiDRA concept can be as simple or as labor- and track-intensive as you like. Simple=2 signal tracks and just a few switching events. Complex=multiple signal tracks and rigorously-enforced level-dependent switching. (The most practical implementation would probably automate the process using computer logic to set and execute level- and time-based switching thresholds.)
Again, I'm happy to agree that better machines with updated technologies are a good thing. HiDRA is a control-layer concept that can be superimposed on any machine layer.
Several people have mentioned in this thread that improved, customized, etc. tape machines exist that do a fantastic job. The implication seems to be that their usable dynamic range is enough, and no one need want for more. I have no problem with that stance if it works for you.
I'm not saying HiDRA would be for everyone. I'm just wondering if it would be an interesting tool for a few wild-eyed adventurers who insist on pushing the analog envelope.
Last edited:
Interesting concept, but at some level I question the need for such a wide dynamic range in practice. At sane listening levels with peaks near 100dBspl at the listening position 60dB dynamic range puts the low end of the range of near the threshold of audibility (or below it) in the typical domestic environment whilst the high end of the dynamic range is probably uncomfortably loud for some range of listeners.
I like R2R analog tape, and use 2 track stereo at 7.5 and 15ips and find it adequate in my listening environment. I haven't done any live recording in recent years so in terms of improved snr on wide dynamic range material this might be a useful technique. I would be more likely to use a peak limiter.. (ouch)
There are very few new machines being made these days, as far as I know the only company still making analog recording gear is OTARI.. I think in so far as this idea is concerned you have missed the boat temporally speaking.
The earlier comment about summing uncorrelated noise (gaussian distribution - i.e random) was actually wrong. The more noise sources you sum together the lower the overall noise. Every time you double the number of noise sources summed you get a 3dB reduction in noise power. (That's precisely why multiple transistors or tubes are paralleled in the front end of low noise amplifiers.) However these different level tracks aren't really summing except when dynamic range expansion is required and then you effectively have modulation of the noise floor which might or probably might not be an issue.
I like R2R analog tape, and use 2 track stereo at 7.5 and 15ips and find it adequate in my listening environment. I haven't done any live recording in recent years so in terms of improved snr on wide dynamic range material this might be a useful technique. I would be more likely to use a peak limiter.. (ouch)
There are very few new machines being made these days, as far as I know the only company still making analog recording gear is OTARI.. I think in so far as this idea is concerned you have missed the boat temporally speaking.
The earlier comment about summing uncorrelated noise (gaussian distribution - i.e random) was actually wrong. The more noise sources you sum together the lower the overall noise. Every time you double the number of noise sources summed you get a 3dB reduction in noise power. (That's precisely why multiple transistors or tubes are paralleled in the front end of low noise amplifiers.) However these different level tracks aren't really summing except when dynamic range expansion is required and then you effectively have modulation of the noise floor which might or probably might not be an issue.
Well it might actually just even out with summing in theory. It just at least seems intuitive to me that you can not solve the s/n problems of one electronic signal by using 3 copies of the same electronic signal. With every added step you need more electronic components which will lead to loss and more noise. Every thing he needs to do with the signal requires an extra step somewhere as far as I can tell. So you are not just comparing 1 signal vs 3 summed in my false premise but you are comparing 2 entirely different signal paths. As far as addressing the switching or fading between the channels I suspect there would be a weird continuity errors or modulation as you put it between the different captures.
Kevin,
I appreciate your insightful comments.
Again, I'm happy to let others make the call for themselves about what dynamic range they want from their analog source. Coming more from the speaker end of things, I'll be the first to admit that source dynamic range is almost never the limiting factor in the playback chain--speaker dynamic range absolutely is the bottleneck, and the ambient noise floor is quite high in most listening rooms.
True, but I'm not silly enough to hope HiDRA will spark an industry renaissance in tape machines. I'm just picturing the HiDRA control layer as a potentially useful way to put these old technological dinosaurs back to work in some niche markets and squeeze the last drop of goodness out of an otherwise obsolete format.
I appreciate your insightful comments.
Again, I'm happy to let others make the call for themselves about what dynamic range they want from their analog source. Coming more from the speaker end of things, I'll be the first to admit that source dynamic range is almost never the limiting factor in the playback chain--speaker dynamic range absolutely is the bottleneck, and the ambient noise floor is quite high in most listening rooms.
True, but I'm not silly enough to hope HiDRA will spark an industry renaissance in tape machines. I'm just picturing the HiDRA control layer as a potentially useful way to put these old technological dinosaurs back to work in some niche markets and squeeze the last drop of goodness out of an otherwise obsolete format.
This is a crucial question that can only be answered by actually trying it.
I'm no expert in human auditory perception, but I think I've read that our ability to register audible events diminishes rapidly with shorter event durations. So, assuming we execute HiDRA's channel switching badly and end up with brief switching artifacts, their perception may be masked more than you might think.
Overall, the proof will be in the pudding. Are there switching artifacts? Does a dynamically adjustable noise floor improve or detract from the listening experience? When all the factors are added together, is the net result of a HiDRA recording better, the same, or worse than the same event recorded in the traditional way? Should we just forget analog altogether and embrace digital wholeheartedly? 🙂 These are the unknowns going forward.
Last edited:
Circling back for a moment to the user-friendliness question, HiDRA could provide recording technicians one very unique and valuable user-friendly feature: it could eliminate the need for re-takes.
In a traditional tape-recording scenario, the technician must exercise extreme care to set levels such that the recording maximizes small-signal fidelity while keeping the peaks from going into saturation. If the musical performance surprises with a higher peak than expected, the only remedy is a re-take. And that's no fun at all.
In a HiDRA recording scenario, the technician can set levels less precisely, allowing plenty of headroom without having to worry about a loss of fidelity.
In a traditional tape-recording scenario, the technician must exercise extreme care to set levels such that the recording maximizes small-signal fidelity while keeping the peaks from going into saturation. If the musical performance surprises with a higher peak than expected, the only remedy is a re-take. And that's no fun at all.
In a HiDRA recording scenario, the technician can set levels less precisely, allowing plenty of headroom without having to worry about a loss of fidelity.
If think I understand the concept, but how fast are you switching tracks? Every time the instantaneous level goes above or below a treshold? (or i cant see it working: some sounds like drums are very peaky so the switch will happen either half way up the leading edge and very quickly (yuck?) ,or in between drum hits (then its just like a gate (recording studio gate)). There will be switching artifacts (how audible is the question)
Most of the recording industry has! And its not just quality but the pain of aligning and maintaining expensive tape machines and paying $300 for 20 min of multitrack tape (which you would triple). Plus tape shed, print thru, climate controlled storage and probably a couple things I missed.
BTW some people still use tape for drums because of the way they peak limit/compress, and record every thing else digitally.
Most of the recording industry has! And its not just quality but the pain of aligning and maintaining expensive tape machines and paying $300 for 20 min of multitrack tape (which you would triple). Plus tape shed, print thru, climate controlled storage and probably a couple things I missed.
BTW some people still use tape for drums because of the way they peak limit/compress, and record every thing else digitally.
- Status
- Not open for further replies.