Topic: Which mobile recorder? M10, H6, or R05?

[Ozpeter]

Some devices are even worse; I have a "24 bit" USB interface that only manages dynamic range of 84dB, because of its analog noise floor.  There is absolutely zero benefit to recording in 24 bit from that device.

In terms of noise, agreed, but isn't there at least a theoretical advantage that using 24 bits helps avoid digital clipping (assuming that you'd not already clipped the analog stage anyway...)?

[Argitoth]

hey tapers, just got my M10. this thread shall not die just yet! Once I get my stuff from Church Audio (two weeks from now?) I'll want to upload A/B comparisons. With external preamp, without, with external mics, without, compare my microphonemadness mics with church audio mics, etc. I want to study my equipment till I'm blue in the face if I can gather enough energy. This will help me decide my next purchase, which I will make some time this year, hopefully no more than 6 months from now.

[F.O.Bean]

hey tapers, just got my M10. this thread shall not die just yet! Once I get my stuff from Church Audio (two weeks from now?) I'll want to upload A/B comparisons. With external preamp, without, with external mics, without, compare my microphonemadness mics with church audio mics, etc. I want to study my equipment till I'm blue in the face if I can gather enough energy. This will help me decide my next purchase, which I will make some time this year, hopefully no more than 6 months from now.

I think you'll be very happy with the CA>M10 rig I ran that as my second rig at second stages at festies and it usually produced great results. They ALWAYS sounded better than you'd think for a $500 setup, that's for sure I even let a friend patch his Schoeps source with my ca14 source once, and we could NOT hear where the patch was once it was edited. Sounds crazy, but its true! You may not find the need to spend a ton of money to upgrade if you're happy with what you're pulling with the ca14s I ran CA14o>9100>m10 and mk41>Littlebox>m10 for the Pimps Of Joytime back in February 2013 and the clear winner of those two sources was the ca14 omnis And it was in a small/crowded bar/venue that was pretty chatty. It was definitely not the most ideal setting to run omnis, but alas, it was the clear winner

[Gutbucket]

Some devices are even worse; I have a "24 bit" USB interface that only manages dynamic range of 84dB, because of its analog noise floor.  There is absolutely zero benefit to recording in 24 bit from that device.

In terms of noise, agreed, but isn't there at least a theoretical advantage that using 24 bits helps avoid digital clipping (assuming that you'd not already clipped the analog stage anyway...)?

24bit makes a larger available range on the digital storage side which will always be considerably more than absolutely necessary to fit the entire range of analog input, but you still need to make sure that smaller range is placed comfortably within that larger range.  Regardless of bit-depth, in either case you need to make sure the input level is adjusted appropriately so the signal fits within in the range between clipping and noise, determined by the entire signal chain.  Too high a input level and the signal will clip in 24bit the same as it does with 16 bit.  Adjusted properly, excess range is split between the headroom up top and footroom(?) at the bottom.  24bit provides additional excess-room over 16bit.  The equivalent of say 14bits worth of actual signal range in a real-world signal will fit within either one.

[bombdiggity]

I even let a friend patch his Schoeps source with my ca14 source once, and we could NOT hear where the patch was once it was edited.

{Posting this "Bean-style"  : }

That's the scary part of it once you drop the big dollars   

It all depends on what you're recording and how though.   

Sometimes the heat is just a waste of money.     Other times there is a clear difference.   

In the OP's case I think higher end gear may be warranted (due to the self noise and quiet source concerns).  OTOH I think he'll be quite happy with his starting point.   

As to the "buy the best don't upgrade" thing that depends too.  As OP notes getting that baseline knowledge is helpful (and many may be satisfied with the right inexpensive choices for their uses).  Also if you buy the right lower end items at a good price there may not be much lost selling them in the YS later. 

[2manyrocks]

All true

At least the OP has a reference point of known gear going forward. 

[Argitoth]

RECORDING: swinging for fun - swinging a usb cable around, with M10 built in mics and iZotope noise reduction, this is just for fun, don't take this recording seriously for any purposes. I then used this same cable to retrieve the recording, for poetic justice.

[DSatz]

argitoth wrote:

> If you're recording sound for use as sound effects/foley stuff, you'll want to record at 24-bit or else you will may hit the 16-bit noise floor when using noise reduction software

If the noise floor of a recorder is the same in 24-bit mode as it is in 16-bit mode, then this objection goes out the window. The content of the 8 low-order bits is essentially random, so your DSP will work harder to process this random data and you will use more bits of storage all to no useful effect whatsoever. Despite the 24-bit internal representation of the signal, its noise is still at the 16-bit level, and will still build up in subsequent processing just as it would with a 16-bit input signal.

When any real-world signal goes through any real-world channel, there is some noise in the signal AND the channel. If either noise level is distinctly greater than the other one, it will predominate; if the two noise levels (and their frequency and time-domain qualities) are similar, they'll combine into something that's a little greater than either one was originally. But in no case does a quiet signal ever make a noisy channel quieter, nor can a quiet channel make a noisy signal quieter. If God created a perfect, noiseless A/D, a digital storage or transmission medium with infinite bits and a perfect, noiseless D/A on the other side, an 80 dB signal that passes through that channel would still have only 80 dB of dynamic range when it came out. The holy miracle of that signal path would consist in the fact that no noise was _added_ to the signal, but it still wouldn't come out any quieter than it was going in.

--best regards

[Argitoth]

argitoth wrote:

> If you're recording sound for use as sound effects/foley stuff, you'll want to record at 24-bit or else you will may hit the 16-bit noise floor when using noise reduction software

If the noise floor of a recorder is the same in 24-bit mode as it is in 16-bit mode, then this objection goes out the window.

yes, the [analog] noise floor is the same at 16 or 24 bit, what is variable is the pre amp gain. you turn down the gain, turn down the noise, and of course the signal (so you potentially get a really low signal, but at 24 bit, that low signal could be perfectly adequate)

http://www.ti.com/lit/ds/sbas288j/sbas288j.pdf
23 bits of noise free resolution for this particular a/d converter.

therefore I don't see how 24 bit does not offer advantages when you fail to set your gain correctly.

[Argitoth]

so who was it that decided to slap on "records at 24 bit" on a mobile recorder that can't achieve such dynamic range (rhetorical question)? So the question is: Does the M10 24 bit mode give any more dynamic range?

[Gutbucket]

That extra bit or two can be enough to be useful when setting input levels in the unpredictable real world, so recording 24bit files on those machines can still be worthwhile.   Just realize that even though the file format is storing 24bit data, you aren't actually recording or storing anywhere close to the maximum dynamic range a 24bit file is capable of.  The extra bit or two is good enough for me to fully justify using 24bit for some of my live recordings.  For consistency and workflow, I simply record 24bit files for all my live recordings.  I would not feel any need to use 24bit for something like LP or analog cassette transfers though, since the levels from those sources are completely predictable and can be fit within a 16bit range comfortably without the need for as much headroom or noise-floor clearance excess.

[Ozpeter]

That extra bit or two can be enough to be useful when setting input levels in the unpredictable real world, so recording 24bit files on those machines can still be worthwhile.

I think what is being said in preceding posts is that if you do that, while there's not much of a problem on the digital side, you are still going to have the traditional problem of an inferior signal to noise ratio produced by under-driving the analog stage and later amplifying it digitally.  I don't think there's an easy answer to the old problem of whether you record at the highest level to get the best signal to noise ratio from the whole system, but at risk of not getting it right and ending up with some clipping (either digital or analog or both).

In the early days of DAT machines with a readout of headroom, I used to love seeing whether I could get to the end of a classical concert not having touched the input level, and with the headroom showing something like -0.5dB. 

If working with familiar equipment, I have a theory that levels are pretty consistent at the microphone (at least with acoustic music).  If recording full orchestra, you're going to have the mics set back somewhat (assuming it's a stereo pair).  If recording an acoustic guitar, you'll be much closer.  The level at the mic will tend to be pretty similar, so your level setting at the recorder will also be much the same.  So at least in that scenario, you can aim a bit higher than -12dB below full scale.

[DSatz]

Argitoth, OK, taking your scenario where you under-record by a significant amount such as 6 or 10 dB--what I'm saying should be fairly obvious, which is that more bits will help ONLY if those bits aren't filled with noise. Extra bits are of use only when the actual noise floor of the recorder (via the inputs and level settings that you're actually using) is lower than that of the signal. Otherwise the content of those bits will be random and meaningless--no better than having a dithered 16-bit recording.

I did test the M10's dither, by the way. It is definitely adequate; there is no "digital deafness" and no observable quantization distortion as a tone is reduced in level approaching (and then going beneath) the noise floor. The problem was just that when the recorder was set to 24 bits, its noise floor decreased by only a smidgen--about 1 dB as I recall--below its dithered 16-bit level.

BTW Jon S., I keep meaning to note that even the simplest form of dither doesn't need to consume the whole dynamic range of the lowest significant bit. Depending on statistical considerations it can be adequate at 1/3 to 1/2 of that 6 dB figure in terms of effective noise power. So the dynamic range of a properly dithered 16-bit recorder should be a few dB greater than 90.

--best regards

P.S.: I still think that people commonly overestimate--perhaps by as much as 20 or 30 dB--the dynamic range that they actually need for live recording in public venues. As long as you can capture the highest peaks of a performance without overload, and play it back without hearing any noise that the recorder itself has added, then you have enough bits! Having one or two bits more, for level-setting comfort so that you don't have to risk overload, is a plus as well.

But what I'm saying is, for nearly all live recording of musical performances that only brings you to about 14 bits total, even including a "comfort" bit or so. I'm fanatical about quiet recordings, and for years in the 1970s I toted 15 ips open-reel recorders and Dolby A-type studio noise reduction systems to record live concerts. You sure as hell couldn't hear any tape hiss in those recordings even when they were played back good and LOUD. But even a dithered 14-bit PCM recorder has a distinctly lower noise floor. For many years the BBC used 12-bit studio-to-transmitter links (although I believe some pre-emphasis was used, which is almost like having maybe one additional bit), and no one heard noise in their famous live FM concert broadcasts.

[Gutbucket]

These bits>

Test your device and learn what its dynamic range actually is, and how much analog gain you need with your microphones and recorders so that the total noise floor does not degrade the mics' dynamic range.  That gain figure is the maximum gain you ever need use (you will use less on a loud source).  Once you know that, leaving enough headroom becomes a simple task.

As long as you can capture the highest peaks of a performance without overload, and play it back without hearing any noise that the recorder itself has added, then you have enough bits!

[Argitoth]

got my church audio mic and preamp. will post some tests soon.