A: The 32 bit IEEE float is a professional format. When using this format you avoid introducing quantization noise. When using a PCM format such as the 16 bit integer format, samples are rounded into 16 bit integer data. By rounding the sample data, a difference between the original data (floating point data) and the data saved in the PCM file may (or rather will) occur - it is the difference between these two signals which is the quantization noise. Quantization noise is exactly that: Noise! and as such it is undesired. Further this type of noise may sound quite nasty (include harmonics). When creating multi source auralisation, many auralisation files are mixed together and this will accumulate noise - including the quantization noise in PCM files. If mixing two files with an equal level of noise - then the level of noise is doubled (+3dB), if mixing 10 files then the noise level increases by 10 dB and if mixing 100 files then the noise level increases by 20 dB. By using the float format at least the quantization noise is avoided.