FFXIV's sound is played on top of a very complicated sound system. In order to implement the orchestrion, we could not play the audio in the same manner as the normal background music due to size restrictions within the system. This meant that we had to decide between not implementing the orchestrion at all or take out various aspects of the system in order to get it into the game. This time we chose the latter option. The reason that the sounds is played using 3D direction audio isn't because we wanted to implement it that way, but because of that fact that it was the best solution available when considering methods to improve the audio quality within the system restrictions.
In other words, one of the things we had to drop in order to implement the orchestrion was the feature which everyone has suggested - the ability to listen to the music in the same quality throughout the house instead of the 3D directional audio.
Let me explain a bit more about why we had to make this call.
There are many different songs in the game, and in order to have the music change instantly, and to be listened by many players at once, we were required to lower the amount of data for each song in the orchestrion. To be more specific, we needed to make the data smaller, approximately 1/50th of a CD quality song in order to implement this. To better explain this as musical data by converting this to measures, say we wanted to play a 5 minute song, but we only have room for 6 seconds but we still needed to make it work.
There are three major ways to lower the sound data:
- Use a high compression codec
- Reduce the sampling rate or bit rate
- Remove the channels
For the first method, this is already implemented in every bit of sound data since the release of FFXIV, so we implemented the second and third option as well. Lower the quality of the audio to the lowest quality, and dropping the channels from being stereo to mono. Honestly, as a sound director, I feel bad that we have to deliver the background music in this quality, and I was thinking of methods to somehow make improvements when implementing it. And this is where we decided to go with the 3D directional audio heard from a furnishing to try and eliminate the audio quality as much as possible. From a data standpoint, this was the only available method to implement, so we made improvements on how the audio is played to get this implemented.
And this is why the system utilizes a feature where the audio is played through a furnishing. We decided that if it's setup in a way that the audio is coming out of a furnishing, even if the quality of the audio is the lowest, and also being played in mono, we can somehow bring it the the level where we can play the songs, and this is how we were able to prepare 47 tracks for release. This is what went on behind the scenes for implementing the orchestrion.
Details on the drastic improvements for the sound driver and data which I mentioned at the start is much more complicated, so I would like to write this up at another time!