We are about to develop a realistic train simulator so its sound should be realistic (not a game for everybody, for only machinists). We'll use FMOD library for sound processing. It has lots of capabilities and works fine. But my question is how to collect real sounds.
We have opportunities to record on a train but we don't know how to collect sound in a professional way analyze them.
Main sound sources are:
• Two train engine sounds in different locations,
• 3 different brake system sounds
• The environment and rail sounds
• Wind whistle
• Lots of fans(ventilators)
• Warning sounds
• Other trains
The first problem is we can't collect sounds in separate parts. At the same time engine is running and railway sound is heard. So engine and railway sound is overlapped. As I know, to separate sounds, one should make n channel recording where n is the sound source number. In a train there are 5-6 sound sources and that is too costly. After the separation in matlab program, the sound quality decreases. How are you doing sound collection?
Secondly, which hardware is used to collect sounds? There are lots of kinds of recorders which differ in quality and price. What do you advice for recorder and recording methods?
Another issue is dynamic sounds such as engines. Engines are affected by engine rpm and consumed current (as the velocity increases or). How can we analyze the sound behavior on these effects so that we can integrate with fmod.
Also the train is interacting with the environment. There are tunnels, bridges, bends etc. In simulators how do you collect such sounds and bind them to fmod.
Not sure if this would be of any help, but Python has a sound module that can do a lot of what you are talking about. I forget the name of the module, but it is the multimedia section of the documentation, under library reference.