How to parse multiline with awk without weird duplicates

Could you please try following.

awk 'BEGIN{  OFS="\t"}/^Name/{  if(value){     print value  }  sub(/.*: /,"")  value=$0  next}/^Description/{  sub(/.*: /,"")  value=(value?value OFS:"")$0}END{  if(value){    print value  }}'  Input_file

You were close:

$ awk -F': ' '/^Name/ {n=$2} /^Desc/ {print n "\t" $2}' filezvbi    VBI capture and decoding libraryzziplib A lightweight library that offers the ability to easily extract data from files archived in a single zip file

The main problem with your script was that the {print...} block was being executed for every line of input rather than just when the Descline was seen and then you weren't including the blank after the : in your FS so it was still present in each field.

And I've written written this sed script:

pacman -Qi | sed -n '/^\(Name\|Description\)[[:space:]]*: /{s///;H}; /^$/ba; $ba; d; :a;x;s/^\n//;s/\n/\t/;p'

1. /^\(Name\|Description\)[[:space:]]*: /{s///;H}; /^$/ba - Each line that starts with Name and Description has the macthed part removed and is appneded to hold space.
2. /^$/ba; $ba; d; - If an empty line or an end of file is encountered, we branch to label a; Otherwise we start new cycle.
3. :a;x;s/^\n//;s/\n/\t/;p - In label a we exchange hold space with pattern space, remove leading newline (dunno where it's from), substitute first newline with a tab, and print the output.

Sample output:

zlib    Compression library implementing the deflate compression method found in gzip and PKZIPzsh A very advanced and programmable command interpreter (shell) for UNIXzstd    Zstandard - Fast real-time compression algorithmzvbi    VBI capture and decoding libraryzziplib A lightweight library that offers the ability to easily extract data from files archived in a single zip file