Hi!
What do you mean by "walking"? 
Is it the bare movement (the opposite of standing still)?
Is it that a person comes closer to an object?
Is it that a person does pass a special spot?
Is it weight on a special spot, because a person steps on it?
You can use different sensors - you have many options.
You could sense movement with a PIR sensor (example: PIR (motion) sensor : ID 189 : $9.95 : Adafruit Industries, Unique & fun DIY electronics and kits).
It is not too selective but it can cover a relatively wide area (maybe 10 square meters).
You can use an infrared proximity sensor (example: Infrared Proximity Sensor Long Range - Sharp GP2Y0A02YK0F - SEN-08958 - SparkFun Electronics)
it is more selective, it will work up to about 150 centimeters.
You can make an infrared light barrier. If a person "breaks" it, the sound will start. It can be pretty selective because the person must pass by a special spot.
You can use a force sensitive resistor (example: Force Sensitive Resistor - Square - SEN-09376 - SparkFun Electronics).
When a person steps on it, it can trigger the sound. Very selective.
These are only some ideas. There are more, different trigger methods.
And you can use combinations of sensors.
How critical is timing?
How long is the sound/music (seconds, minutes... decades)? (see: ASAP)
Is it always the same sound(file) or different ones?
Example 1 (own project)
Some time ago I was asked to help with some "sound trigger installation" for the exhibition of a friend who makes sculptures.
Only one or two sculptures were in a small room (about 4-5 square meters), there were about 8 of this small rooms (in total about 12 sculptures). He wanted a sound to start playing, when a person enters each of the small rooms. The sound-file was 1 out of 5 to 10 files (some kind of randomness), about 30 seconds to 1 minute.
The artist said: "Some seconds after a person does enter the room the sound should start to play - one or two seconds more or less is OK. After the person (or all persons if there are more) left the room and no one else is there, the track shall play until its end and stop. After that there should be silence until an other person enters the room."
I decided to use PIR sensors. And they did the job pretty well.
Other hardware used (main parts):
Arduino UNOs
DFPlayers (Serial mp3-player with micro sd card)
Audio amplifiers
Loudspeakers
Example 2 (own project)
We had about 10 TV sets (kind of) within an exhibition. The curator wanted that every TV set should play a short movie while a person is standing right in front of the TV set. The movies (soft sound only) were about 20 to 30 seconds long. The TV sets were placed side by side with about 2 meters distance from each other.
I used (long distance) infrared proximity sensor placed just below each TV set. I set the trigger distance to about 100 to 150 centimeters. This made the desired effect.
Other hardware used (main parts):
Arduino MICROs
Laptops with video-player
TV sets (loudspeaker integrated)