How to read different bits from a binary file?

Hi. I am trying to read a binary file with MATLAB which contains different bits (8-bits and12-bits).

I read stream 1 easily with ('*uint8'). But can you help me to read stream 2 and 3 from the binary file, please?

Answer 1

[Sulaymon Eshkabilov]

Just employ: 'unit16' instead of 'uint8'.

Answer 2

[Walter Roberson]

I could read stream 1 (8-bits) with a command of "stream1 = fread(fileID, 113:3600112 '*uint8')".

No, do not do that. Instead fseek forward by 112 bytes from the beginning of the file, and fread with size 3600000 with *uint8.

For the second group you do not need to move after the first group. fread 3600000 elements using 'ubit12=>uint16'

For the third group you do not need to move after reading the second group. Do the same fread 3600000 'ubit12=>uint16'

I was concerned because the majority of time that data structures are spoken about as 12 bit, really what is meant is using 16 bits per sample with either the 4 MSB or LSB unused. However the sizes work out perfectly for it to really be 12 bits per sample.

[Walter Roberson]

There is a tip for location streams. As I mentioned stream 2 starts at 3600112 and this means that the file is automatically read from the continuation and ends in 3600112 + 3600000 = 72001112. But stream 3 does not start from this point and starts from 9000112!

Stream 2 starts from byte 3600112 . When you read multiple items using ubit12, then MATLAB reads in bytes but breaks them up at the bit level, so when you ask to read 3600000 items, it would read 3600000 * (12/8) = 5400000 bytes. And 3600112 + 540000 = 900112.

filename = tempname();
fid = fopen(filename, 'w');
fwrite(fid, 0:111, 'uint8');
after_header_pos = ftell(fid)

after_header_pos = 112

fwrite(fid, 0:35, 'uint8');
after_stream1_pos = ftell(fid)

after_stream1_pos = 148

stream1_length_bytes = after_stream1_pos - after_header_pos

stream1_length_bytes = 36

fwrite(fid, uint16(256+(0:35)), 'ubit12');
after_stream2_pos = ftell(fid)

after_stream2_pos = 202

stream2_length_bytes = after_stream2_pos - after_stream1_pos

stream2_length_bytes = 54

fwrite(fid, uint16(4095-(0:35)), 'ubit12');
after_stream3_pos = ftell(fid)

after_stream3_pos = 256

stream3_length_bytes = after_stream3_pos - after_stream2_pos

stream3_length_bytes = 54

fclose(fid);
dinfo = dir(filename)

dinfo = struct with fields:
       name: 'tpb72ec826_d22b_4f99_a138_245ed8e9d2df'
     folder: '/tmp'
       date: '28-Nov-2020 19:57:19'
      bytes: 256
      isdir: 0
    datenum: 7.3812e+05

dinfo.bytes

ans = 256

delete(filename)

Observe how I wrote 36 ubit12 items, but that the space used on the disk was 36*(12/8) = 54 bytes.

[Walter Roberson]

Which sample file are you referring to?

hlen = 112;
s1len = 3600000;
s2len = 3600000;
s3len = 3600000;

headerdata = randi([0 255], 1, hlen);
stream1_data = randi([0 255], 1, s1len);
stream2_data = randi([0 4095], 1, s2len);
stream3_data = randi([0 4095], 1, s3len);

filename = tempname();

%write phase. Get some data into a file with the needed structure
fid = fopen(filename, 'w');
fwrite(fid, headerdata, 'uint8');
after_header_pos = ftell(fid)

after_header_pos = 112

fwrite(fid, stream1_data, 'uint8');
after_stream1_pos = ftell(fid)

after_stream1_pos = 3600112

stream1_length_bytes = after_stream1_pos - after_header_pos

stream1_length_bytes = 3600000

fwrite(fid, stream2_data, 'ubit12');
after_stream2_pos = ftell(fid)

after_stream2_pos = 9000112

stream2_length_bytes = after_stream2_pos - after_stream1_pos

stream2_length_bytes = 5400000

fwrite(fid, stream3_data, 'ubit12');
after_stream3_pos = ftell(fid)

after_stream3_pos = 14400112

stream3_length_bytes = after_stream3_pos - after_stream2_pos

stream3_length_bytes = 5400000

fclose(fid);
dinfo = dir(filename)

dinfo = struct with fields:
       name: 'tp0872a127_ae91_4228_8107_b9dab80ae7c0'
     folder: '/tmp'
       date: '28-Nov-2020 22:31:21'
      bytes: 14400112
      isdir: 0
    datenum: 7.3812e+05

dinfo.bytes

ans = 14400112

%read phase
fid = fopen(filename, 'r');
headerdata_in = fread(fid, [1 hlen], '*uint8');
stream1_data_in = fread(fid, [1 s1len], '*uint8');
stream2_data_in = fread(fid, [1 s2len], 'ubit12=>uint16');
stream3_data_in = fread(fid, [1 s3len], 'ubit12=>uint16');
fclose(fid)

ans = 0

delete(filename)

if isequal(headerdata, headerdata_in)
    fprintf('header data all read in okay\n');
else
    fprintf('header data mismatch. Wrote\n');
    disp(headerdata);
    fprintf('received\n');
    disp(headerdata_in);
end

header data all read in okay

if isequal(stream1_data, stream1_data_in)
    fprintf('stream1 data all read in okay\n');
else
    fprintf('stream1 data mismatch. Wrote\n');
    disp(stream1_data);
    fprintf('received\n');
    disp(stream1_data_in);
end

stream1 data all read in okay

if isequal(stream2_data, stream2_data_in)
    fprintf('stream2 data all read in okay\n');
else
    fprintf('stream2 data mismatch. Wrote\n');
    disp(stream2_data);
    fprintf('received\n');
    disp(stream2_data_in);
end

stream2 data all read in okay

if isequal(stream3_data, stream3_data_in)
    fprintf('stream3 data all read in okay\n');
else
    fprintf('stream3 data mismatch. Wrote\n');
    disp(stream3_data);
    fprintf('received\n');
    disp(stream3_data_in);
end

stream3 data all read in okay

[Eliza]

Thanks a lot. Your answer was so helpful.