How to convert signed integer (from -8 to 7) into 4 bit signed binary

I wanted to convert a vector (approximately 270k entries) of integers ranged from -8 to 7 into 4 bit signed binary, and I used

dec2bin((typecast(int8(Data), 'uint8')), 8)

to find the 8 bit signed binary for my data. However I need to take either the upper 4 bits of the 8 bits, or convert them into 4 bit signed binary directly from the integers as they can be expressed as 2's complement 4 bit signed binary. Could anyone teach me how to do that?

My current situation: I have all of them as 8 bit signed binary, but some entries only appear as '10', '1', '11', etc. i.e. not 8 bits even though I tried to use

dec2bin(d, n)

So when I tried to convert the numbers to strings and take the first 4 bits with

num2str

all I would get are spaces.

Any help would be greatly appreciated!!

Answer 1

[Walter Roberson]

int2bin4 = dec2bin([15:-1:8,0:7],4) - '0';   %or leave out the '0' if you want character output;

translated_vector = int2bin4(YourVector(:) + 9, :);

[Jeffrey Chen]

Thank you, could you explain what "int2bin4" does? I typed the first row and it gave me the signed binary of -8 to 7, and I tried to replace "YourVector" with my data it gave me the "Subscript indices must either be real positive integers or logicals." error.

[Walter Roberson]

What is min() of your data? Your question says that it is -8 but the error message implies that it might be more negative.

Or possibly there are non-integer values or NaN or inf values:

nnz(isnan(YourVector))
nnz(isinf(YourVector))
nnz(YourVector ~= floor(YourVector))

All three of those should come out as 0 if the data is in the range you indicated.

[Jeffrey Chen]

All three gave me 0, but I just realized the max and min are 15 and -13 respectively... so there must be a mistake when I'm normalizing the decimal value from -1 to 1 to integer values.

My goal is to take that vector of about 270k elements generated by Audacity that range from -1 to 1 (which should be 16 bits) and convert (or normalize) them into 4 bit signed binary values directly, or 8/16bit and take the 4 most significant bits. Do you have any suggestions for to me to approach this? My original command to get (what I thought would be) -8 to 7 integers was

Data_int = round(Data_original.*(2^4-1)./max(Data_original));

but apparently the max/min elements turn out to be more than what I expected. Thank you so much!!

[Walter Roberson]

Data_int = int2bin4(floor(Data_original * 15/2) + 9, :) ;

int2bin4 is acting as a lookup table to provide an efficient way to do the data conversion