CELT TODO: Difference between revisions

From XiphWiki
Jump to navigation Jump to search
No edit summary
No edit summary
Line 18: Line 18:
** Partial remixing of short blocks?
** Partial remixing of short blocks?


* Make split more efficient
<strike>* Make split more efficient</strike>
** Support for LN=-1
<strike>** Support for LN=-1</strike>


* Minimise bits left after PVQ
* Minimise bits left after PVQ
Line 32: Line 32:


== Other decisions to make ==
== Other decisions to make ==
* Should we use raw bits or not?
<strike>* Should we use raw bits or not?</strike>
* What should we do about remaining bits after the last PVQ?
* What should we do about remaining bits after the last PVQ?
** more fine energy?
** more fine energy?
Line 38: Line 38:
** allow "fractional pulses"
** allow "fractional pulses"
* Include a configuration packet for more flexibility? (probably not)
* Include a configuration packet for more flexibility? (probably not)
* Use a min width for ebands that's equal to the number of short blocks?
<strike>* Use a min width for ebands that's equal to the number of short blocks?</strike>

Revision as of 08:01, 25 May 2010

Open issues

  • Tuning
    • Band layout (spacing)
    • Bit allocation
      • Main table
      • Stereo
      • qtheta offset
      • fine offset
      • energy prediction parameters (means, decay)
    • Spreading/folding parameter
  • Transients
    • Haar/Hadamard
    • Delta adjustment
    • Can we remove all of the old code?
    • Split below 32-bit
    • Partial remixing of short blocks?

* Make split more efficient ** Support for LN=-1

  • Minimise bits left after PVQ
    • Reverse band ordering?
    • Fractional pulses?
  • Dynamic allocation
    • Include a way to deviate from the normal bit allocation?
  • Better stereo coupling
    • Better support for "intensity stereo"?

Other decisions to make

* Should we use raw bits or not?

  • What should we do about remaining bits after the last PVQ?
    • more fine energy?
    • encode bands backward?
    • allow "fractional pulses"
  • Include a configuration packet for more flexibility? (probably not)

* Use a min width for ebands that's equal to the number of short blocks?