ruby, `match': invalid byte sequence in UTF-8

I have some problem with UTF-8 conding. I have read some posts here but still it does not work properly somehow.

That is my code:

#!/bin/env ruby
#encoding: utf-8

def determine
  file=File.open("/home/lala.txt")          
  file.each do |line|           
    puts(line)
    type = line.match(/DOG/)
    puts('aaaaa')

    if type != nil 
      puts(type[0])
      break
    end        

  end
end

That are the first 3 lines of my file :

;?lalalalal60000065535-1362490443-0000006334-0000018467-0000000041en-lalalalallalalalalalalalaln Cell Generation
text/lalalalala1.0.0.1515
text/lalalala�DOG

When I run this code it shows me an error exactly when reading the third line of the file (where the word dog stands):

;?lalalalal60000065535-1362490443-0000006334-0000018467-0000000041en-lalalalallalalalalalalalaln Cell Generation
aaaaa

text/lalalalala1.0.0.1515
aaaaa

text/lalalala�DOG
/home/kik/Desktop/determine2.rb:16:in `match': invalid byte sequence in UTF-8 (ArgumentError)

BUT: if I run just a a determine function with the following content:

#!/bin/env ruby
#encoding: utf-8

    def determine
    type="text/lalalala�DOG".match(/DOG/)
    puts(type)
end

it works perfectly.

What is going wrong there? Thanks in advance!

EDIT: The third line in the file is:

text/lalalal»DOG

BUT when I print the thirf line of the file in ruby it shows up like:

text/lalalala�DOG

EDIT2:

This format was also developed to support localization. Strings stored within the file are stored as 2 byte UNICODE characters.The format of the file is a binary file with data stored in network byte order (big-endian format).