Audio input, quality hardware?

10s

10s

Veteran
I know the answer might be (drum roll)...NAB, :dankk2:but I'll try anyway. Maybe there's an already existing answer I haven't seen... or maybe there's a better answer than just .....NAB, afterall a forum where most answers end with NAB is .....pretty DRAB.

Will this camera have audio inputs given the dockable nature of it's configuration and if so, will it incorporate pro sample rates up to 24bit or better i.e., 96k sample rates or even the memory hog 192k? How about excellent pre-amps?

Video is great because it incorporates sync audio and gives us 48k sampling, better than CD! unfortunately most mic preamps in video cameras are garbage and unsuitable for pro usage, the DVX was an exception and now Canon and others are cleaning up their act, literally. So given that Jannard comes from my neck of the woods, WA state and given that we have some decent sound design/mfg firms here in Seattle ( Mackie, Rane...) it would make sense to have Excellent pre amps designed & built into the camera by knowledable design professionals (not accountants). The typical cheap stuff slapped on as an after thought would simply degrade the tool. The good stuff would make a huge splash in the audio world and really help indie storytellers take their work to a even higher level.

Any answers? i can hear the ducks already...nab nab nab nabnab
 
10s said:
Will this camera have audio inputs given the dockable nature of it's configuration and if so, will it incorporate pro sample rates up to 24bit or better i.e., 96k sample rates or even the memory hog 192k? How about excellent pre-amps?
Click to expand...

That wouldn't be very smart from RED. A decent A/D dsp (that can actually do REAL 24bit / 96khz) cost several thousand $. Most people are happy with dvx/hvx audio and the pros won't record the audio with the camera.
 
Marcus X, you'll have to tell numerous companies that already mfg. DSP/preamp boxes that hook up to computers to stop selling their award winning products ( & they help win Grammys etc... also) for way under the price you listed.

Just yesterday I was looking at Focusrite, the maker of Class A preamps & other signal processing gear, and they have their new product out in affordable terms. I happened to see another ad for building your own top quality prez, etc... in other words everything needed to increase the audio capture ability is currently available at moderate prices.

One reason pros don't record audio "in-camera" is simply because the audio components aren't there in film cameras and in video ( until recently) they weren't the best. Change that and the practice will follow. Remember that back in the 20s there were numerous film/sound formats (Some formats recorded sound on the side) but the system that won out was the dual (current) system because of editing needs and many practicle considerations. The new generation of video challenges all that. The old audio tradition continues, but that should not suggest that it is the only course of action nor the best , just as film is being questioned as the only alternative for feature productions. I happen to use a dual system sometimes with my 96k 24 bit flash recorder and then other times I use the cameras built in 48k 16 bit system. It works well. I would hope a new camera design would incorporate what my little 96k 24 bit flash recorder has in it and go even further & beef up the preamps, that's where miking coloration is best handled before entering the A/D coverters and DSP. Thats where people notice and say, that sounds really good!
 
The problem with 24 bit audio is the same as HVX vs. "real 1080".

Go to a computer shop and tell them you need something to record 24 bit audio. They will sell you a PCI soundcard for $50, which records 24 bit audio files, supports 3d audio, digital and analog output and some games are included.

Then you go to a big music store that sells studio equipment and ask the same thing. They will show you some "inexpensive" simple A/D racks, 2 channels, just input/output and a volume-knob that starts at $1,500 like the one from Apogee.
When you tell them you can get a soundcard for $50, they'll start talking about dB, SNR, noisefloor, etc.

Hook up your flash-recorder to an oscilloscope and you'll find out it's actually something like 11-13 bits. The 24 bits are just "uprezzed".

16 bit is enough for Red.
 
So Marcus,

Can you explain some more on what you know about flash-recorders. Also, what do you think of the Tascam HD P2? Is that a true 24 bit?
 
I googled for the tascam and they claim it has a dynamic range of 105 dB, which means it's 18 bits max. But you won't need more unless you want to record a starting jet or a rocket ;)
 
Marcus, I am still curious about what you actually mean. Can you refer me to an article or maybe even explain this to someone who is not an expert with sound. And is DV true 16bit?

thanks a lot
 
I'm not an expert either, but I have some synthesizers and other audio gear and I searched for the best way to record the audio. I looked for A/D converters and found out there are alot of cards/racks which are 24 bit/96 kHz between 50$ and 5000$.
I wanted to know what the difference is and its not about features, but about dynamic range and signal-to-noise-ratio (SNR). This is what the quality of audio depends on.

My Soundblaster Live card (16 bit) has about 80 dB. Sounds good, but that means that about 20% of the 16 bit audio is just noise. So you can call it a 13 bit card, or "not a real 16 bit" card. For 16 bits, noisefree audio you need about 100 dB.

The audio file is just a container (like .avi, .mov, .mxf) for signal + noise. To get the best quality, you have to minimize the noise and maximize the signal.

Some newer cards and chips have 110 dB, up to 120 dB. So 16 bits is not sufficient to record every information. But it would be silly to produce "17 bit, 18 bit, 21 bit, ..." converters. The next step would be 20, or 24 bits.
Most manufacturers use 24 bit resolution when the dynamic range is above 100 dB. But that does not mean they have a 24 bit signal. They actually waste about 6 bits with noise.
The higher the dynamic range, the more expensive the A/D converters are. To reduce the thermal noise, you have to install some serious active cooling devices.

About DV: DV is 12 or 16 bits. It's just the container. To find out how good your camera records the audio, you have to ask the manufacturer about dynamic range and SNR. Sometimes they print this in the manual. I just looked at my DVX100AE manual, but it just says "12/16 bit".

But the big question is: who needs anything above 16 bit / 100 dB?

- Music producers need that as a headroom for mixing or alternating/pitching/bending/transformating the signal.
- People recording extremly loud stuff (way louder than a rock concert).
- Scientists

For just recording and playback real-world stuff, you won't need that. Listen to a good CD, that's "oldschool 16 bit".
 
Hi,

I think I can help clarifying the issue...
Here goes:

most Big-Budget film production record their production-Sound (on-location-dialog) on stereo DAT tapes. these record 16bit/48kHz, however, the DAT-Decks are about $6000 machines with superb A/D converters. When using top-quality microphones and pre-amplifiers and then converting to digital with top quality A/D converters, 16bit/48kHz is more than enough for dialog and production-sound in general. The extra bits will not make such a notable difference. (could make a very slight difference though...)
On the contrary, you can get a cheep microphone, connect it a to a cheep pre-amplifier that is likely included in a cheep PCI card with low-end A/D converter that claims to do 24bit/96kHz and guess what? it will sound very bad, and will create files about 4 times bigger, making it much heavier on computer and storage recourses.

Conclusion:

16bit/48kHz with top-notch pre-amps and converters is the industry standard.
24bit/48kHz has a much better dynamic range, therefore is superior.
24bit/96kHz is simply more than you need for production recording.

However, 24/96 should be used as a master delivery format, since it does have advantages for music and sound-effects, and enable a more dynamic and crispier sounding final mix. 24/96 is also the delivery format for DVDs.
 
Last edited:
Good points about the need for more than 16 bits. A good CD-player can get more than 100 dB signal to noise ratio from a good CD, which has a theoretical limit of 16*6= 96 dB S/N (dithering tricks gets more s/n out of it). At home even 100dB is LOUD. I can pump out about 110 dB continuous 18-20kHz from my 1KW stereo system, but that is unbearable! And there is always at least 40 dB noise background in every house, which gives a usable S/N of only 60 dB in real life...

And the fact is that we can not hear a difference between 16/48k and 24/96k as 16/44.1k is just about the limit of human hearing resolution. Thus the final carrier needs not be any better, even if higher bit depths and sample rates do give some safety and quality advantages to audio engineers.

4 channels of 24/48* would be nice for amateurs using RED (???), but pros will continue to record audio in Devas, Cantars etc and on-camera audio is not really needed.

*) not for "quality" but safer level setting.

PS 24 bit recorders of course ARE 24 bit recorders, but very few have noiseless preamps enough to utilise that huge S/N ratio. Is like buying a 4 gallon bucket with 1 gallon of dried cement in everyone of them. Just to clarify (?) MarcusX's point above.
 
Last edited:
Waaaah! You're mixing up all kinds of decibels!

dB of dynamic range have nothing to do with dB of playback volume.

24 bit recordings have advantages other than increased dynamic range. Dudes... the increase in dynamic range is indeed often unimportant in most recording situations. That's because the range is added to the QUIET end of the dynamic range, where the noise floor is. So discussing the benefits of -100 vs -102 dB in a typical 30db background noise environment is iffy at best, except in a good studio.

However, 24 bits provides 256 times the RESOLUTION of 16 bits, and that's where the benefit really lies. The sound is smoother. And this allows you to be a bit more conservative with your levels - if you record at 24 bit with the meters only reaching half of the possible dynamic range, you're still getting over 100 times the resolution of 16 bits. So, in fact, the benefits people are touting as "extended dynamic range" is really the opposite - "the ability to get more out of a smaller dynamic range".
 
Not so, adding more bit depth (16 -> 24) only gives more dynamic range. Each bit gives 6 dB more dr. Threre is no more "resolution". Absolutelly not. It is not the same thing as moving from SD to HDTV.

And loudness and dynamnic range are related. If you play "perfect" 16 bit recording so that the loudest passages are 100 dB in loudness (forget different weightings for now...) the backround noise is 4 db in theory. If you crank up the volume to 120 dB (deaf...) the noise floor is now 24 dB (can not hear it even now for obvious reasons). With 24 bit recording you can play at 144 dB before the noise gets above 0dB (limit of hearing). This is theoretical, of course.
 
No dude, sorry, it ain't so. The resolution difference is tremendous, and all that mostly matters. Try and find yourself a chart of how the bits break down. Anyway, playback volume measurements are in dBSPL; dBu and dBv are used for analog voltages; and dbFS is for digital signals relative to full-scale (0). These are all different, and involve conversion factors depending upon everything from gain controls to equipment calibration to digital manipulation.

Anyway, each extra bit doubles resolution. It does NOT double dynamic range. dB's are logarithmic measurements. The difference between -96 and -95 dB is tiny compared to the difference between -1 and 0 dB. Look at your meters... see how the db values are so close together near the low end, and far apart near the high end? That's the logorithmic scale.

Every digital recording device can record up to 0dBFS, or "full scale". The analag front end determines where this is, and thus where peaking will occur.

A 16-bit recording will get you (theoretically) down to about -96 dBFS. 24 bits will get you to about -144dbFS.

In almost any situation, including a really good sound-isolated studio, you have ambient noise of at least 20 dBSPL. A typical set/quiet location will be about 40dbSPL. The threshold of hearing damage is around 120 dbSPL. Typical "outrageously loud" situations (e.g., loud concerts) max out at arround 110dbSPL.

So the range of ACTUAL SOUND in most shooting situations is much less than that of 16-bit samples.

The additional dynamic range says nothing of how loud the sounds are that you can record - you determine that with your gain settings; it'll be 0dfBFS. The additional dynamic range of 24 over 16 bits comes in below -96 dbFS. And there's generally nothing useful there, as we've seen.

So let's say we're recording converational speech, which tends to run around 60dbSPL. We set our gain so that this equals, say, -12dbFS. Then, our dynamic range in dbSPL goes from about 0 to 72dbSPL in either 16 or 24 bit (since SPL is always positive). 24-bit goes a tad closer to 0 (in the range between 0 and 1 dB), but with the noise floor above 20dB, it's not really a big deal.

Now, do you think that anyone would really build a system in such a way that the 256-fold increase in resolution between 16 and 24 bit sampling will affect only the bottom end of a system? the space between 0 and 1 dB? No. Quite unreasonable. The 24-bit system has increased resolution throughout the dynamic range. The space between -1 and 0dB, much more important than the space between -143 and -144 db, has a lot more resolution in a 24-bit system. In the useful range of 60 or so dB for most applications, 24 bit recordings have much more resolution than 16-bit systems. And THAT is the real important difference.
 
Last edited:
Sorry Ullanta, I hope somebody with better english than me cuts in here to explain what logarithmic scale means, how it relates to digital recording, why adding bit depth only adds dynamic range, no "resolution" etc...

In many ways your eplanations of level settings are correct, but there is something in your basic undestanding of the backround mathematics that is wrong (logarithmics and resolution).

About level meters: the dB:s near the top are more apart because we want the ballistics of the meters that way; we like to know the top end more accuratelly than the low end, which is really not interesting at all. Thus meters are doubly logarithmic, from the dB-system itself, and from the design of the meters. A 1 dB difference is a 1 dB difference is it between -90 and -91 or between -2 and -1 dB.
 
Sorry Petrus (this is fun).

In the decibal world, every 3 db increase is a signal about twice as (loud, strong, forceful).

0 dbFS is as loud as it getrs.

-3 dbfs is half as loud

-6 dbfs is half as loud as that, or 1/4 as loud as 0

-9 dbfs is half as loud as that, or 1/8 as loud as 0

-12 dbfs is half as loud as that, or 1/16 as loud as 0.

....
-27 dbfs is 1/512 as loud as 0.

-30 dbfs is 1/1024 as loud as 0.

I could go on, but indeed the difference between 1/512 (-27 dBFS) and 1/1024 (-30 dBFS) is 1/512 of the level at 0 dBFS. The difference between 1/2 (-3 dbFS) and 0 (0 dBFS) is 1/2 of the level at 0dBFS. The three-decibel range from -30 to -27 dbFS is 256 times smaller than the range from -3 to 0 dBFS. So, in overall terms, a 1 dB difference is a RELATIVE difference, not absolute - it does not always mean the same thing. 1 dB less than -90 dB is a much smaller difference than 1 dB less than 6 dB, just like 20% of 10 is much less than 20% of 1000.
 
Last edited:
The last message addresses the concept of the logarithmic decibel scale. Now, for resolution:

16 bits = 65,536 steps of resolution; or ~682 steps/dB average over 96 dB

24 bits = 16,777,216 steps of resolution; or ~116,508 steps/dB average over 144 dB

That's A LOT MORE RESOLUTION! See?
 
Last edited:
Avarage resolution has no place here, each relative difference gets the same amount of bits: 6dB difference is allocated one bit and that's it. Adding bits just stretches the available scale.

Othervise no objections to your dB 101. The reason why logarithmic scale is used is that the precieved loudness difference between, say -54 to -60dBF is the same as the difference between -6 and 0 dBFS. In absolute terms the energy difference is what you say, but it is always only the relative difference we are interested in. Thus logarihmic scales.

I forget what we are arguing here about??? Yes, adding bits (= 6dBFS each bit) we just stretch the range in which we can capture audio without noise. Basically what we gain is that we can push the noise floor lower down. It does not make the range more finely devided (more resolution). I repat myself.

An analogy: do pianists, it they want more "resolution" play only on left side of the keyboard, as the note values down there are at closer absolute intervals? Would addind more keys to the end of the keyboard make the notes more finelly devided in the old octaves?

Yes, this is fun!

(As a pro photographer/engineer I understand log scales perfectly well, thank you)
 
Last edited:
Petrus said:
I forget what we are arguing here about??? Yes, adding bits (= 6dBFS each bit) we just stretch the range in which we can capture audio without noise. Basically what we gain is that we can push the noise floor lower down. It does not make the range more finely devided (more resolution).
Click to expand...

You're talking about the digital noise floor, which has nothing to do with the noise floor (in dBSPL) out in the world. Best to keep clear on the distinction. Except in the most perfect of situations, the "real world" noise is orders of magnitude higher than the digital noise floor. Now, think binarily:

Adding one bit to a binary number doubles the range of possible values. By definition, this doubles the resolution. Since the HIGH END of the value range is fixed (0 dBFS), this has the effect of letting us register additional info on the bottom end... that is, we can now measure with a granularity half the size. This extends dynamic range by definition - the tiny little ratio on the bottom is being extended. Every time we increase precision, we by definition increase dynamic range on the bottom end- since we can distinguish a smaller "smallest value".

(As a pro photographer/engineer I understand log scales perfectly well, thank you)
Click to expand...

Are we throwing credentials around?
 
You must log in or register to reply here.
Back
Top