json.dumps() is what you want here, if you use print(json.dumps(pairs)) you will get your expected output:

>>> pairs = {'arun': 'maya', 'bill': 'samantha', 'jack': 'ilena', 'hari': 'aradhana'}
>>> print(pairs)
{'arun': 'maya', 'bill': 'samantha', 'jack': 'ilena', 'hari': 'aradhana'}
>>> import json
>>> print(json.dumps(pairs))
{"arun": "maya", "bill": "samantha", "jack": "ilena", "hari": "aradhana"}

Here’s a basic print version:

>>> print '{%s}' % ', '.join(['"%s": "%s"' % (k, v) for k, v in pairs.items()])
{"arun": "maya", "bill": "samantha", "jack": "ilena", "hari": "aradhana"}

# do not use this until you understand it
import json

class doubleQuoteDict(dict):
    def __str__(self):
        return json.dumps(self)

    def __repr__(self):
        return json.dumps(self)

couples = [
           ['jack', 'ilena'], 
           ['arun', 'maya'], 
           ['hari', 'aradhana'], 
           ['bill', 'samantha']]
pairs = doubleQuoteDict(couples)
print pairs

Yields:

{"arun": "maya", "bill": "samantha", "jack": "ilena", "hari": "aradhana"}

The premise of the question is wrong:

I know, json.dumps(pairs) does the job, but the dictionary 
as a whole is converted into a string which isn't what I am expecting.

You should be expecting a conversion to a string. All “print” does is convert an object to a string and send it to standard output.

When Python sees:

print somedict

What it really does is:

sys.stdout.write(somedict.__str__())
sys.stdout.write('n')

As you can see, the dict is always converted to a string (afterall a string is the only datatype you can send to a file such as stdout).

Controlling the conversion to a string can be done either by defining __str__ for an object (as the other respondents have done) or by calling a pretty printing function such as json.dumps(). Although both ways have the same effect of creating a string to be printed, the latter technique has many advantages (you don’t have to create a new object, it recursively applies to nested data, it is standard, it is written in C for speed, and it is already well tested).

The postscript still misses the point:

P.S.: Is there an alternate way to do this with using json, since I am
dealing with nested dictionaries.

Why work so hard to avoid the json module? Pretty much any solution to the problem of printing nested dictionaries with double quotes will re-invent what json.dumps() already does.

The problem that has gotten me multiple times is when loading a json file.

import json
with open('json_test.json', 'r') as f:
    data = json.load(f)
    print(type(data), data)
    json_string = json.dumps(data)
    print(json_string)

I accidentally pass data to some function that wants a json string and I get the error that single quote is not valid json. I recheck the input json file and see the double quotes and then scratch my head for a minute.

The problem is that data is a dict not a string, but when Python converts it for you it is NOT valid json.

<class 'dict'> {'bill': 'samantha', 'jack': 'ilena', 'hari': 'aradhana', 'arun': 'maya'}
{"bill": "samantha", "jack": "ilena", "hari": "aradhana", "arun": "maya"}

If the json is valid and the dict does not need processing before conversion to string, just load as string does the trick.

with open('json_test.json', 'r') as f:
    json_string = f.read()
    print(json_string)

It’s Easy just 2 steps

step1:converting your dict to list

step2:iterate your list and convert as json .

For better understanding check down below snippet

import json
couples = [
               ['jack', 'ilena'], 
               ['arun', 'maya'], 
               ['hari', 'aradhana'], 
               ['bill', 'samantha']]
pairs = [dict(couples)]#converting your dict to list
print(pairs)

#iterate ur list and convert as json
for x in pairs:
    print("n after converting: nt",json.dumps(x))#json like structure