Skip to main content

Topic: Floating points? (Read 1745 times) previous topic - next topic

0 Members and 1 Guest are viewing this topic.
Floating points?
In reading up on the differences between Flac and Wavpack, one major difference is that wavpack can handle floating point bitrates and flac can not. 

When would one want floating point bit rates? Would it ever be used in playback?
Music lover and recovering high end audiophile

  • spoon
  • [*][*][*][*][*]
  • Administrator
Floating points?
Reply #1
You would not use on playback, however you might use it on production where it eliminates the possibility of clipping.

  • Kohlrabi
  • [*][*][*][*][*]
  • Global Moderator
Floating points?
Reply #2
Not the bit rate is floating point, but the bit depth is, that is, the bits per sample. Generally, the more bits you have for your bit depth, the higher the resolution of your signal (though floating point works a little bit different). I'm a bit short on time, so I will just refer you to the wikipedia page about bit depth.
  • Last Edit: 20 July, 2015, 07:51:49 AM by Kohlrabi
It's only audiophile if it's inconvenient.

Floating points?
Reply #3
As spoon's pointed out, it makes more sense using it from the audio engineer's POV*, such as this guy's:

https://www.youtube.com/watch?v=92CJdNJen9E

You can (if memory serves me well), cut to the chase to about 6min into it, but to make sense out of it, you'd better watch the whole thing, him dithering or not.


*IMO, not too different from high bit depths and sampling rates.
  • Last Edit: 20 July, 2015, 09:23:21 AM by includemeout
Listen to the music, not the media.

Floating points?
Reply #4
Besides the threads dealing with floating vs. fixed point maths are so numerous on HA, you're bound to find one or another post/reply that can expand into it.
Listen to the music, not the media.

Floating points?
Reply #5
Thanks for the information folks. I appreciate it.
Music lover and recovering high end audiophile

  • bryant
  • [*][*][*][*][*]
  • Developer (Donating)
Floating points?
Reply #6
Many, if not most, DAW (Digital Audio Workstation) packages use floating-point data natively, for lots of good reasons. They generally store this data in WAV files (or WAV extensions like BWF), but WavPack provides a compressed alternative, and this can be especially attractive because on a percentage basis 32-bit float data compresses very well (better than 24-bit does) and these applications often have tracks with lots of silence, which of course compresses to essentially nothing.

Among the leading DAWs, Cockos's Reaper and Steinberg's WaveLab both support WavPack natively, and Adobe's CoolEdit / Audition support WavPack with plugins that I provide.

But, as was said above, this really does not have a lot of application for most users, as most people would not have access to anything but fixed-point integer or DSD audio.