With the same hardware and playback scenario (local file, streaming, etc), any OS has the equal potential to sound the best. At the end of the day, it's about getting the PCM audio data to the buffer used by the soundcard/device. What makes Windows more challenging is how the audio application communicates to the sound card/device through a virtual driver. In other words, the audio playback application has to go through a middle man (universal audio device) when using MS approved drivers. What's even more challenging is how other applications can impact the middle man, just by starting an application or updating the version can cause custom settings to change.
Windows 8 has introduced an audio hardware acceleration path which enables custom drivers or even the application to gain exclusive access to the hardware, but has limited hardware support.
With that said, Mac OS and Linux can have all this same exact issues, just with linux, you become your own boss on how to configure and update the OS and applications and if you don't like the options available, you can change the configuration or write a little C/C++ to create it.
At the end of the day, you should be able to configure Windows, Mac OS and Linux to all be bit-perfect, but in a consumer product, i would want complete control when it comes down to reliability of playback.
Jim