Hiya Markus, >Basically, you should record everything as loud as possible, then >change the volume in your sampler (or logic mixer). Yep, that's good basic practice, so long as you're not getting digital overs / clipping. Though in a 16 bit system there are actually 65,536 sample values available, ranging from -32768 to +32767. >A good pre-amp with a compressor / limiter function will allow you to >get a hot level into the digital domain without having to leave excess >headroom for dealing with transient attacks, and the end result is >that the signal will sound better, and use more of the dynamic range >available to it from the bit-depth that you're working with.. More detail in the signal, probably the same dynamic range. The compression / limiting reduces dynamic range above threshold - levels are then turned up for better resolution - this gives extra dynamic range below the threshold. >The same goes for normalisation - record loud, then do your >normalisation. >If it changes it more than a few dB, you didn't record it loud enough in >the first place, and should go back to the source. Normalization, as a destructive process with a 16-bit output, degrades signal quality somewhat since in most software there is no dithering. For best sound quality the signals should not be normalized, but amplified within the floating-point format of Logic's audio mixer, and finally dithered at the master fader. Cheers, Thomas