# Difference between revisions of "CELT TODO"

Jump to navigation
Jump to search

Line 6: | Line 6: | ||

** Decide on folding gain | ** Decide on folding gain | ||

* dynamic allocation | * dynamic allocation | ||

− | * Using PVQ for coarse energy/ | + | * <strike>Using PVQ for coarse energy?</strike> |

* configuration packet | * configuration packet | ||

* revisit stereo coupling | * revisit stereo coupling | ||

* stereo allocation | * stereo allocation | ||

* tune everything | * tune everything | ||

− | * increase max pulses | + | * <strike>increase max pulses</strike> |

− | ** more pulses for N>5 | + | ** <strike>more pulses for N>5</strike> |

** <strike>Fast CWRS for N=3</strike> | ** <strike>Fast CWRS for N=3</strike> | ||

** <strike>change if(_n<=6) to if(_n<=6 || _m>255) in ncwrs_u32</strike> | ** <strike>change if(_n<=6) to if(_n<=6 || _m>255) in ncwrs_u32</strike> |

## Revision as of 05:31, 15 June 2009

## TODO list for CELT (before freezing)

- raw bits
- Use the last bits after PVQ (for fine energy?)
- adaptive folding
- finalize folding details
- Decide on folding gain

- dynamic allocation
~~Using PVQ for coarse energy?~~- configuration packet
- revisit stereo coupling
- stereo allocation
- tune everything
~~increase max pulses~~~~more pulses for N>5~~~~Fast CWRS for N=3~~~~change if(_n<=6) to if(_n<=6 || _m>255) in ncwrs_u32~~~~Use real limits in fits_in32()~~

- Tune for lower sampling rates
~~Improve dynamic range by changing the energy floor~~~~Laplace encoder~~~~Make ft=32768 to save some divisions~~~~Use the last symbols to have a wider range~~

- Investigate other transient algorithms
~~energy weight~~