The Middle Ages

From the late 1990s game music moved to recorded tracks on CD. But memory based audio systems were not capable of reproducing full length high quality music. At this time soundtracks were stored on the CD media and played back in exactly the same way as a regular CD during play. This is the time when so called "audio engines" began to appear. An audio engine, as the term is used today, is really an elaborate buffering and memory management system. Early audio engines did not cater for streaming music so the playback of game music relied on a direct connection between the CD-ROM and the soundcard. Now, a powerful game console or PC can load several minutes of audio into memory in a second, replay hundreds of channels simultaneously from RAM or stream and mix many audio channels from secondary storage using DMA. This is partly due to the advantages of compressed audio like FLAC, MP3 and Vorbis and hardware decompression, but primarily its a simple function of improvements in system power. Presently, a game sound system has the capabilities of a professional sampler from the turn of the century. It can play back hundreds of simultaneous voices, apply envelopes, filtering, panning and surround sound localisation, modulation and reverb to the stored samples. But game sound is still about 5 or 6 years behind the state of the art in DSP. Although technologies like Puredata and Csound have been around for a long time this level of DSP capability has not yet been adopted as standard in game sound. Tools like FMOD [30] and Wwise [1] are presently audio delivery frameworks for pre-recorded data rather than real ``sound engines" capable of computing sources from coherent models. This is starting to change now, and it brings the history of game audio full cycle.