32 bit float or 48 bit.īecause I hesitate between sequoia, pyramix and protools hd for mixing.
#Bottleneck waveburner free#
As the reference point moves around to utilise a greater overall dynamic range, the noise floor moves around with it, so you don't get a free lunch and gain staging is still worth paying attention to.įor you which is best calculation to have a beautiful image stereo and a beautiful definition.
#Bottleneck waveburner 32 bit#
However, with a 32 bit float system, you still only have a signal that has an instantaneous 144 dB dynamic range (24 bits for the signal, 8 bit mantissa). As far as internal processing goes, if you're using floating point, you can scale the gain all over the place and have very hot signals that won't overload until it comes time to output at 24 bits to the outside world. In the floating point world, 32 bits is single, 64 is double.Įach bit is still 6.02 dB, but you can resolve a much smaller signal with twice as many bits. This is commonly referred to as double precision. 144 dB dynamic range becomes 288 dB dynamic range. 24 bits times two, or doubled, equals 48 bits. Double-precision processing, as I understand it, applies only to internal sample rates. Am I wrong?Īll of this is great, but none of it says "double precision" to me with respect to bit depth. For summing stages, 8 additional bits are used to add 48 db of headroom (hence the 56-bit label), and the data is never truncated to 24 bits until it hits a physical output (including a bounce-to-disk).Īll of this is great, but none of it says "double precision" to me with respect to bit depth. Once inside the system, further calculations are truncated at 48-bits at every stage. Rather than rounding off at 24 or 32 bits, as a floating-point system would, the entire 48-bit value enters the TDM mixer. Accorrding to the white paper, initial "input" to the mixer stage is based on multiplying the 24-bit input times a 24-bit value representing pan and gain, producing a 48-bit mathematical result. The 48-bit TDM mixer explicitly has a 288 db dyanamic range, end-to-end. There may be such a system, but I've never heard of it. "Double-precision" as you used the term would seem to suggest that a 48-bit system is encoding 6 db over 2 bits, double the absolute resolution of a 24-bit system, while maintaining the same dynamic range of 144 db. What I think matters is single or double precision, not the numbers 24, 48, 32 or 64.Can you explain that a little? I'm not aware that any PCM system is double-precision with respect to bit depth - that is, 6 db of dynamic range is always encoded into one bit of data. 64 bit float and 48 bit fixed are both double precision. 32 bit float and 24 bit fixed are both single precision.
0 kommentar(er)
