how to normalize a value to a range between 0 and 1

Question

Hi - I'm doing an audio experiment where recording the sound coming out of my speakers gives me a number between around 0 and 20. I want to convert this number to something between 0 and 1, so that I can translate the volume level gets translated to the scale of an object. How do I do this? My basic setup is:

 var audioVal:float = audio.GetOutPutData(...); //gives the audio in decibels
 var scaleVal:float; //this value will give a scale between 0 and 1;
 scaleVal = Mathf.Clamp(audioVal,0,1); //this doesn't seem to work. The value always ends up being at the maximum or minimum end of the scale.

Answer 1

Best Answer

Answer by Eric5h5 · Jun 27, 2012 at 07:25 PM

Just divide by 20, or you can use Mathf.InverseLerp for arbitrary ranges. Clamp won't work because it just clamps numbers, it doesn't try to scale anything.

Add comment · Show 2 · Share

heaversm · Jun 27, 2012 at 07:40 PM 0

Share

Thanks, Lerp should work great.

meat5000 ♦ · Jul 23, 2015 at 11:58 AM 0

Share

http://answers.unity3d.com/questions/1012209/algorithm-help.html

Similar question, answered with Clamp.

Answer 2

Answer by hideouswill · Jun 27, 2012 at 08:40 PM

Mathf.Clamp() does not scale a value; it restricts the value to the range. For example, if you clamp between (0, 1), any value greater than 1 will yield a clamped value of 1, and any value less than zero will yield zero; for a value inside the clamp range, the value will be unchanged.

To scale, you need to divide your raw value by the total range, and account for an offset if min != 0. For a range of (min, max):

 scaledValue = (rawValue - min) / (max - min);

For the common case where min == 0:

 scaledValue = rawValue / max;

Add comment · Show 7 · Share

TechSupportIncoming1 · Jul 23, 2015 at 09:42 AM 0

Share

In a case where max could be theoretically infinite, should I just use an arbitrary big number as a max value?

saschandroid · Jul 23, 2015 at 09:57 AM 1

Share

c#:

 float max = float.$$anonymous$$axValue;

Bunny83 · Jul 23, 2015 at 11:22 AM 2

Share

@TechSupportInco$$anonymous$$g1:
It makes no sense to map an unbounded value to a bounded range. If "1.0" should represent infinity and "0.0" represent "0", then all values would mathematically map to "0.0".

Think about a real world example: temperature. The temperature has a lower bound of 0 kelvin (-273.15 °C) but is unbounded at the top. Why would you want to map an arbitrary temperature into the range of [0.0, 1.0]. All commonly used temperatures would be zero or extremely close to zero. So even temperatures around 1 million kelvin would result to be almost 0.

In such cases you might use a non linear mapping like (1/x) or log():

 v1 = 1f - 1f / (1f+t);

That's the mapping of the first values:

 t       v1        v2
 -----------------------------
 0       0         0
 1       0.5       0.00990
 2       0.66666   0.01960
 3       0.75      0.02912
 4       0.8       0.03846
 5       0.83333   0,04761
 6       0.85714   0.05660
 42      0.97674   0.29577
 100     0.99009   0.5
 333     0.99699   0.76905
 1234    0.99919   0.92503
 +inf    1.0       1.0

As you can see the values between 0 - 1 are mapped to 0-0.5 and the values between 1 - infinity is mapped to 0.5-1.0. If you need a better resolution for values around 100 you can either divide t by 100 beforehead or use 100f ins$$anonymous$$d of 1f:

 v2 = 1f - 100f / (100f+t)

This will make 100 to map to 0.5

Here's a google link to that formula. You should see a graph at the top. make sure you zoom appropriately.

meat5000 ♦ · Jul 23, 2015 at 11:33 AM 0

Share

It makes no sense to map an unbounded value to a bounded range

(You might need to in a shader)

TechSupportIncoming1 · Jul 29, 2015 at 10:17 AM 0

Share

@Bunny83 thanks for the elaborate and scientific answer :)

Show more comments

Answer 3

Answer by AnotherYeti · Jun 27, 2012 at 08:40 PM

If the range you're getting is exactly between 0 and 20, then replace your third line with:

 scaleVal = audioVal / 20.0;

The code you're using isn't working because Mathf.Clamp merely restricts values to the range given, rather than doing a transformation from one space to another.

Add comment · Share

Answer 4

Answer by Split3 · Aug 12, 2019 at 04:32 PM

Just in case you want to stop at some point, this code works just perfectly for that!!!

 var scaledValue = Mathf.Clamp(value / maxValue, minValue, maxValue);

Enjoy

Add comment · Show 1 · Share

Bunny83 · Aug 13, 2019 at 11:31 AM 0

Share

This makes no sense. Since "value" is supposed between $$anonymous$$Value and maxValue, dividing value by maxValue gives you a range of 0-1. Clamping that range between $$anonymous$$ and max makes no sense. Also your lower bound would be wrong. You don't seem to be interested in a 0-1 range but in a $$anonymous$$Value to maxValue range which was not asked in the question. Scaling by maxValue makes no sense at all. There are cases where something like this is wanted:

 $$anonymous$$athf.Clamp(value * scale, $$anonymous$$Value, maxValue);

where scale controls the scaling and clamp just limits the outcome to a certain range. However as I said this has nothing to do with the question that was asked.

how to normalize a value to a range between 0 and 1

4 Replies

Your answer

Follow this Question

Related Questions