Room mic a full band

mudslide73

I'm using a Tascam DR05 and a Superlux E523D on a stand. Better than the onboard mics but nice and cheap. Quite an honest representation of what happened at the practice usually.

AntonHunter

fastonebaz said:

Ooh 32 bit floating point recording is excellent technology.   Those new zoom gadgets just need a decent small form factor, low light video camera with 4hr battery on board and every band will buy one. 

The low-light video camera will be the tricky thing!

And does 32 bit float basically mean we don't need to bother setting levels?

fastonebaz

AntonHunter said:

fastonebaz said:

Ooh 32 bit floating point recording is excellent technology.   Those new zoom gadgets just need a decent small form factor, low light video camera with 4hr battery on board and every band will buy one. 

The low-light video camera will be the tricky thing!

And does 32 bit float basically mean we don't need to bother setting levels?

Basically yes.

And apparently at a theoretical level it has capacity to handle upto 1500db, and there isn't a sound on earth that gets near that level so it should be good for most applications.

octatonic

fastonebaz said:

AntonHunter said:

fastonebaz said:

Ooh 32 bit floating point recording is excellent technology.   Those new zoom gadgets just need a decent small form factor, low light video camera with 4hr battery on board and every band will buy one. 

The low-light video camera will be the tricky thing!

And does 32 bit float basically mean we don't need to bother setting levels?

Basically yes.

And apparently at a theoretical level it has capacity to handle upto 1500db, and there isn't a sound on earth that gets near that level so it should be good for most applications.

I don't recommend people get in the habit of thinking this way but you are not wrong.

Just using your post as a launching pad- I imagine you know this but others might not.

Dynamic range is more of a 'down to' thing than an 'up to' thing though.
So a 100dB dynamic rang is the difference from digital zero(the maximum a signal can be before it clips) to where it disappears into the noise floor.

Clipping is greatly reduced with 32-bit floating point vs 24-bit but the part that matters here with adjusting gain levels is the 'floating point' bit.
A 24-bit fixed point dynamic range is 142dB.
A 32-bit fixed point dynamic range is 192dB.
That isn't what allows you to recover audio that appears to be clipped, although it does give you substantially more dynamic range in the first place.
That is mostly down to it being floating point.

Fixed Point:

In fixed-point representation, the fractional part of a number is represented as an integer multiple of a fixed small unit, such as a fraction of a second or a decimal point. This means that the number of digits after the decimal point is fixed and does not change.

Floating Point:

In floating-point representation, the fractional part of a number is represented as a binary fraction with a variable number of digits. This allows for a much wider range of values and greater precision than fixed-point representation. 

1st explanation:
Floating-point numbers have a variable number of digits before and after the decimal point. This means that the number of bits used to represent the number can vary depending on the magnitude of the number. The decimal point can “float” relative to the significant digits of the number. For example, a 32-bit floating-point number can represent numbers with a much larger dynamic range than a 32-bit fixed-point number.

2nd explanation: simpler.
This can, in a fairly reductive way, be explained as giving you the ability to move where the decimal point is.
So if you visibly clip a signal, let's give that a value of 1.0 (where 0.0 is the maximum).

With fixed point you cannot move the decimal point.

With floating point you can so what is 1.0 can become 0.1.

(Maths nerds, yes I know- I am trying to explain it as simply as possible).

3rd explanation: It is maaaaaagic.

It is better though to think that you can get out of trouble if levels are set incorrectly.
Being able to gain stage is still a useful skill.

It just isn't as critical now to set things exactly right.

Stuckfast

Just to add, 32-bit floating point operation doesn't make a device impossible to clip. It just means you can't clip it at the converter. It will still be possible to overload the analogue circuitry with a hot enough signal.

AntonHunter

octatonic said:

fastonebaz said:

AntonHunter said:

fastonebaz said:

Ooh 32 bit floating point recording is excellent technology.   Those new zoom gadgets just need a decent small form factor, low light video camera with 4hr battery on board and every band will buy one. 

The low-light video camera will be the tricky thing!

And does 32 bit float basically mean we don't need to bother setting levels?

Basically yes.

And apparently at a theoretical level it has capacity to handle upto 1500db, and there isn't a sound on earth that gets near that level so it should be good for most applications.

I don't recommend people get in the habit of thinking this way but you are not wrong.

Just using your post as a launching pad- I imagine you know this but others might not.

Dynamic range is more of a 'down to' thing than an 'up to' thing though.
So a 100dB dynamic rang is the difference from digital zero(the maximum a signal can be before it clips) to where it disappears into the noise floor.

Clipping is greatly reduced with 32-bit floating point vs 24-bit but the part that matters here with adjusting gain levels is the 'floating point' bit.
A 24-bit fixed point dynamic range is 142dB.
A 32-bit fixed point dynamic range is 192dB.
That isn't what allows you to recover audio that appears to be clipped, although it does give you substantially more dynamic range in the first place.
That is mostly down to it being floating point.

Fixed Point:

In fixed-point representation, the fractional part of a number is represented as an integer multiple of a fixed small unit, such as a fraction of a second or a decimal point. This means that the number of digits after the decimal point is fixed and does not change.

Floating Point:

In floating-point representation, the fractional part of a number is represented as a binary fraction with a variable number of digits. This allows for a much wider range of values and greater precision than fixed-point representation. 

1st explanation:
Floating-point numbers have a variable number of digits before and after the decimal point. This means that the number of bits used to represent the number can vary depending on the magnitude of the number. The decimal point can “float” relative to the significant digits of the number. For example, a 32-bit floating-point number can represent numbers with a much larger dynamic range than a 32-bit fixed-point number.

2nd explanation: simpler.
This can, in a fairly reductive way, be explained as giving you the ability to move where the decimal point is.
So if you visibly clip a signal, let's give that a value of 1.0 (where 0.0 is the maximum).

With fixed point you cannot move the decimal point.

With floating point you can so what is 1.0 can become 0.1.

(Maths nerds, yes I know- I am trying to explain it as simply as possible).

3rd explanation: It is maaaaaagic.

It is better though to think that you can get out of trouble if levels are set incorrectly.
Being able to gain stage is still a useful skill.

It just isn't as critical now to set things exactly right.

Great explanations! I especially like number 3...

The main reason I was asking is the impression I got is that rehearsal recording, or even most gig recording (my use case) doesn't really reach the dynamic range where 32 bit float would be useful, so long as someone knows how to set gain. More useful in, say, orchestral music or field recording.

My H5n already has a feature that records a backup of the stereo pair at 6db lower which is more than adequate, and I can't remember the last time I needed that.