If you are considering a shotgun/cardioid...

[Alex H.]

But something like RAW where you can to some extent change recording parameters after the fact would probably be a big enough incentive. That would probably "obsolete" your recorder, though it wouldn't change the fact that you could still do great recordings. DAT is still a very acceptable recording quality, but you can't get tapes anymore, at least not reasonably available. So if some better storage comes along and everyone jumps on, what will kill your recorder is not that it doesn't sound up to snuff but that you can't find cards for it any more. Try finding a RLL hard drive these days...

Interesting idea! I'm trying to wrap my brain around what an audio equivalent of RAW might be, and how it would work. Gonna need another cup of coffee.

For storage, it seems that SD is taking over for most things, aside from high-end cameras with proprietary storage media. As SD read/write speeds and storage limits increase, and the prices keep falling, they seem to be flooding in as a standard. Now, will that last? Who knows? Assuming that SD is here to stay would be like your reminiscence of DAT: never assume. But It seems to be the thing for now, and it keeps improving in ways that a format like DAT couldn't.

What I'd be most worried about right now would be wireless. As more of the spectrum faces sell-off and privatization, more of the units many of us own are threatened by outlawed operation. Bluetooth and WiFi digital systems are finally coming into their own, with a quite a way to go yet, but as users increase, just how much saturation can one venue handle?

 

[graywolf]

Actually, I thought wav files were the audio equivalent to raw video/photo files. Most people who do not edit their audio record mp3's.

 

[Noiz2]

WAV et all are lossy formats. They only sample every so often and have no clue what happened in-between. The closest I know of to a RAW like audio format is bitstream, but that never caught on. Nobody who records professionally records to MP3. Consumers record to MP3 a fair amount and a lot of voice recorders record to MP3.

The biggest problem with very lossy formats like MP3 even for consumer purposes is that the encoders get freaked out by some kinds of sounds. Classics in that category are bagpipes and sirens going by, the doppler makes some of them go crazy. You won't get that with WAV because it isn't doing post capture data compression.

 

[ggrantly]

I'm no Ludite but FWIW, I haven't purchased a 4K camera either. It hasn't effected business yet.

Grant

 

[graywolf]

Someone ought to make a sticky of Richard's post, about as accurate as you are going to get in less than 100 pages or so.

I, maybe, should have said that a high bandwidth uncompressed .wav file is the audio equivalent to a raw video file.

I did assume that folks here relialized that chopping analog into digital is never 100% lossless. In fact it is only 50% at best, but the bits you lose can be so small they are beyond human perception.

 

[Noiz2]

If you use that definition of "lossy" then ALL digital formats are "lossy". Audio, Video, Temperature, Pressure, etc. etc.

True, but some are a lot more lossy than others. The point is that WAV is not lossless.

Yes, that is how ALL digital recordings work. If they sampled continuously and infitessimally, they would be analog.
However, sample rates are chosen to be high enough to allow accurate reproduciton of the highest frequency of interest. In audio, that is essentially 20KHz. So we use samples a bit higher than 2x the highest frequency of interest. For example, "CD qualty" uses 44,100 samples per second. And in the video world, we typically use 48K samples/second. This was published by Harry Nyquist in 1928 and further research by Claude E. Shannon in 1948. It is popularly known as the Nyquist-Shannon Sampling Therom.

That is the theory. If it was completely correct then nothing over 44.1/16 would sound different. So it is more like Newtonian Physics, it's good as far as it goes. The problem is that it's based on the limitations of the ear and the ears are not the only way information is taken in and processed by the brain which is where raw data becomes "hearing".

A really obvious example of the lossyness of fixed sample rate formats is the difference between 16 bit and 24 bit which is quite audible by pretty much everyone.

As for "no clue what happened in-between", that is why we sample so frequently, so that nothing unpredictable happens "in-between". And we apply a low-pass filter to the audio (typically a frequency somewhat higher than 20KHz) so that there can't be anything beyond the comprehension of the sampling rate.

Well then we wouldn't need oversampling would we. Depending on the frequency quite a bit can happen between samples. Again nobody would be using anything over 44.1 if nothing was missing between the samples at 44.1. And you are kind of making my point by talking about using filters to toss data so that the signal fit's into the limitations of the converter.

Axtually all digital data is a "bitstream". From the temperature reports from your local airport to 8K video pixels in 24-bit color.

Fun semantics game. "Bitstream" is the name of a recording format, but of course you are right that all digital streams are a stream of bits. Bitstream is a digital format that like RAW is an acquisition format and very data dense and much closer to a true lossless format.

WAV (and AIFF) are actually lossless, at least the way we most commonly use the term. While there are some varieties of WAV that use lossy data compression, it is very rare to see it out in the Real World. Fraunhoffer's MP3 has become the overwhelming default data compressed format in modern times. Of course any digital data can be described as a "bitstream" or a stream of bits.

There are two common uses of the word lossless that apply here and we are using different ones. You are referring to data compression post acquisition, I am talking data loss during acquisition. Yes WAV takes what the converter puts out and leaves it alone. You have me with the compressed WAV file, I have never heard of data compression as part of a WAV file. But we are both talking about the standard PCM WAV file that is common. And again I'm not sure if you have never actually heard of Bitstream or you are playing word games. "Bitstream" is also called "1 bit recording" Korg MR-1000 and Direct Stream etc. The MR-1000 samples at 5.6448 MHz.


I was wrong actually by referencing RAW files since really they are just all the data that the sensor puts out and that is basically the same as what we are getting in a WAV file. I was referring to the post flexibility in the RAW format. But that flexibility is also very dependent on how much data you collect.


The point originally (before the rabbit hole adventure) was that short of a major change in the amount and way audio recorders capture data what you have now will probably work fine. And the more likely thing to orphan a recorder is the demise of the recording media in favor of something new.

 

[Noiz2]

I, maybe, should have said that a high bandwidth uncompressed .wav file is the audio equivalent to a raw video file.

And my use of RAW was off. I really meant a very data rich file with the flexibility in non destructive post processing that RAW files have.

I did assume that folks here relialized that chopping analog into digital is never 100% lossless. In fact it is only 50% at best, but the bits you lose can be so small they are beyond human perception.

It's the "perception" part that is still under debate. But in a broader way I'm talking about that "perception" after altering a file. IF what you record is going to be played as recorded then the bar is lower. If you want to be able to adjust a file and still be clearly above the "perception" level then you need more data to start with.

An example is recording at a high sample rate when you are planing on slowing down a sound. It is very common to slow a sound down by 50% or so in SFX work and if you do that to a 44.1/16 file it sounds bot great. Traditionally the work around was to use some samplers to ingest the sound at a higher SR. Part of why Synclavier's were still in use in the 1990's, they sample at 100K. And why it is not uncommon to record SFX at higher sample rates now.

 

[graywolf]

Well, Scott, there are recorders out there that will do 32bit/192kz. My Zoom F4 will only do 24/192 being a cheap recorder. And, woe on to me, my camcorders will only do 16/48.

But then, my computer only has 32 cores to work with. There always seems to be limitations... Doesn't there?

 

[Noiz2]

There always seems to be limitations... Doesn't there?

Well yes there are. The OP wanted to know if his gear was likely to be outdated soon.

So, I guess to rephrase my answer. When the limitations of your gear are lower than the expectations of your clients, yes it's time to get some new gear.

The rest was my speculation on what might cause that.

If your point is your cores are bigger than mine, well maybe so. I'm happy with the size of my cores even if yours are bigger.

 

[combatentropy]

Not sure if anyone here would actually disagree with what is said in this video. But I like how it says it.

 

[Joshua_G]

Having electronics engineering background, I'm well familiar with the above theories, demonstrations and statements.

However, listening experience defies those theories.
I'm yet to hear a digital sound system that will fool me to think I'm hearing an analog one. (I'm not the only one in this).

 

[graywolf]

If your point is your cores are bigger than mine, well maybe so.

No, my point was exactly what I said. No matter how great your equipment is there are still limitations to what it can do.

 

[Noiz2]

Your link is to data compression loss, and I was very clear that is NOT what I was talking about.

Magnetic tape is also essentially "digital". Magnetic particles are either magnetic or not, they live in one binary state or the other. There is just a LOT of them and the D/A is an averaging converter.

16 bit and 24 bit both top out at 0, there is no "additional headroom". What I assume you are referring to is that if you pretend you are at 16 bit and record lower then you have more "imaginary" headroom. But the digital scale is dBfs and -20 is the same level at 16 bit as it is at 24 bit, both are 20dB below full scale.

Well marketing tends to use all kinds of words that have no relation to the real meaning. Just because you can't seem to grasp a concept doesn't mean it's meaningless. But you do seem to be embracing the "post truth" concept quite well.

The "1 bit" you are talking about is indeed a marketing term, you obviously didn't read or understand what I was talking about which is a very different beast.

Well I'm just going to stop this weird tangent. I don't know if you are being dense or just trolling.

 

[graywolf]

16 bit and 24 bit both top out at 0, there is no "additional headroom". What I assume you are referring to is that if you pretend you are at 16 bit and record lower then you have more "imaginary" headroom. But the digital scale is dBfs and -20 is the same level at 16 bit as it is at 24 bit, both are 20dB below full scale.

I wish I could get a definative answer on that.

There are three ways to see that:
1-- If the A/D sets 16 ones to 10 volts and 24 ones to 10 volts, then zero dbBFS is the same for both, but all the other values vary.
2-- On the other hand most charts seem to indicate that it sets 16 ones to about 6 volts and 24 ones to 10volts. In which case 0 dbFS is different.
3-- The thrid way is if the A/D sets 1 volt to -20 db on the 24 bit scale, and to -12 db on the 16 bit scale, and lets the other values fall where they will.

It may well be that it is a case of whichever the designer of the AD converter chose at the time. The point is that there is no way you can make 8, 16, 24, and 32 bit scales match each other. The can match at the top, at the middle, or at the bottom but not overall.

Since all audio scales cross to zero dbVU, I would guess that is what most if not all designers work to.

Where you are most likely coming from is that most equipment seems to actually be 24 bit today, and that is what is actually being used even when converted to 16 bit by software/firmware. But then how does that compare to older 16 bit equipment?

Confusing!

 

[Noiz2]

Actually I'm coming at it from a digital perspective. What comes out as a voltage is adjustable in the D/A section, but digitally you top out at the same place, you just divide it differently.

It's like you have $1 in change. You are trying to describe somethings value. On one hand you have a bunch of nickels and on the other you have pennies. If what you are describing is worth $0.50 you are good in either case but if it's worth $0.53 then you are going to have round up or down with the nickels. So the nickels (bit depth, number of counting pieces) is less accurate in describing the real worth.

Humans hearing has it's limitations but we are very sensitive to perceived errors. It perks up our defensive alert system.

I haven't seen a real in depth study of perception VS physical limitations, but there was some work done on tone variance perceptabiylity and humans can pick up a tonal difference that is a lot finer than what you can physically "hear" directly. This also works with expectations. If you know, or think you know, what something should sound like you are very sensitive to missed expectations.

Now if we had all set a standard that each digital number would convert to a specific voltage increase, and the analog side ran in a fixed way then you would be getting more volume with higher bit counts. If we did a direct word to output then switching from -20dBfs 16 bit to -20dBfs 24 bit would boost the electrical output by 256 times, since every bit doubles the resolution.
Flash to Back To The Future...

But that is not how we do it. We calibrate the zero to a set voltage, and count down from there. So instead of getting more max output at the same resolution (voltage/ number) we get the same output at a higher resolution.

You don't get more headroom but you get more detain in your headroom. The calibrated analog out put from -20dBfs to 0dBfs stays the same but there are a lot more smaller steps to get there.

 

[graywolf]

It's like you have $1 in change.

As far as I can determine zero-dbFS is what the converter puts out as all ones. That in itself means nothing.

Half that signal would be -6 db. That in itself means nothing.

Half the difference would be -3 db. This does mean something, it means that is the highest value that will not clip. Does not matter if we are measuring peak-to-peak, root-mean-square, or average. Half the signal is above and half is below that -3 db value. If you move it up above that -3 db the upper sideband will clip. Has nothing whatsoever to do with the actual level of the analog signal, but everything to do with the digital signal. OK?

Now if zero-dbVU is defined as -20 dbFS, then zero-dbFS is 10 volts rms. That is because zero-dbVU is defined as one volt rms.

Then the only difference is how low of a signal the A/D converter can measure.

But, generally, in 16 bit digital zero-dbVU is defined as -12 db. This make the two scales entirely different from each other. They are not the same system of measurement. Sort of like degrees F and degrees C.

Why is this all so important. Well, my Zoom F4 has a 24 bit A/D converter, while my Camcorder has a 16 bit A/D converter. Which leads to the question, how do I set values between them?

What I do is set the recorder to -20 db, and the camcorder to -12 db peak reading. Both seem to agree those values are equal to zero-dbVU.

Those three values seem to be the only commonality between my recorder, my camcorder, and my analog mixers.

Yet, you seem to keep insisting that they are all the same?

 

[Noiz2]

?

It seems to me your talking specifics about your brand. Maybe I'm missing something.

IF I play a calibration tone out of my speakers and switch from 16 bit output to 24 bit out put the level doesn't change.

When getting into the theory I'm talking digital. The analog side can be set to just about anything. In the good old days it was, it's why you had to lay calibration tones on your tape.

On a digital scale -20 dBfs is 20 dB from all ones, it's a fixed relationship. dB is a relationship between values.

I'm not doubting that what you do to calibrate between your equipment isn't correct but you are calibrating between the analog side of the D/A converter and the analog side of the A/D converter. It has nothing to do with digital.

It is entirely possible that the two machines don't agree on what is an analog zero, but it doesn't change the bits that came before.

 

[graywolf]

Strangely, Scott, I am not disagreeing with you. I understand the analog side real well, but dBFS came about long after my education. I look on the web for info and get conflicting stuff. I get opinions, which seems to me to be what you are giving as well. I do not find definitive answers.

An interesting factoid. If I set my slate tone for -20db and send it to the camcorder that is what I get. If I set the ST to -12 and send it to the camcorder that is what I get. Only they both are at the same gain setting in the camcorder. Different numbers but the same gain setting?

Makes no sense whatsoever. I am beginning to believe that there is no standard for dBFS at all. That it is just some malarky someone made up.

Sort of like the definitive answers on Wikipedia that any one can change at will.

 

[Noiz2]

That is how it should work. Once you have calibrated them to each other then whatever you send should be what you get at the other side. Maybe I'm misunderstanding what you are doing.

More than opinion, but not the world authority either. Opinion based on lots of reading 30+ years of experience, designing and building both digital systems and analog circuits, including a digital analog hybrid synth in the 80's and having to slog through the transition from all analog to all digital in production facilities. I started as a physics major and spent years doing engineering for a small R&D firm and at one point was doing software testing at a major company. So it's at least informed opinion.

FS just means the top is actually the top. In analog you can keep going, you start to get a lot of distortion but you can keep going. Also in analog zero is where you set it. There are standards but nothing forces you to follow them. In digital zero is where you run out of numbers. There is nothing beyond full scale and FS means the same thing everywhere, the end of the line.

 

[Rick R]

To confuse Tom more, when the first R-DAT machines reared their ugly heads (pun), my fellow (NYC) recording studio engineers and I refereed to -18dBFS as "digital zero" as the DAT was calibrated to the console's meters @ 0VU (using a 1kHz test tone).. Which was +4dBu at the console's XLR outputs .. at least that's what our Gold Line test set stated. When measured with a volt meter, it was 1.2 volts AC as I recall.