Allowing much deeper band splits#17
Conversation
jmvalin
force-pushed
the
jmvalin/deep_band_split2
branch
from
6bab893 to
c65b6bd
Compare
jmvalin
force-pushed
the
jmvalin/bits2pulses2
branch
from
0093a99 to
bed775f
Compare
jmvalin
force-pushed
the
jmvalin/deep_band_split2
branch
from
c65b6bd to
d6879fd
Compare
Removes the LN=-1 split limitation and changes the band layout so that all bands can now split all the way down to N=2 or N=3.
jmvalin
force-pushed
the
jmvalin/deep_band_split2
branch
from
d6879fd to
5965c49
Compare
jmvalin
force-pushed
the
jmvalin/bits2pulses2
branch
from
bed775f to
7348434
Compare
|
|
||
| static const oac_int16 eband5ms[] = { | ||
| /*0 200 400 600 800 1k 1.2 1.4 1.6 2k 2.4 2.8 3.2 4k 4.8 5.6 6.8 8k 9.6 12k 15.6 */ | ||
| 0, 1, 2, 3, 4, 5, 6, 7, 8, 10, 12, 14, 16, 20, 24, 28, 34, 40, 48, 60, 78, 100 |
There was a problem hiding this comment.
Does this fix a long-standing typo or is there an advantage to switching to a 16-bin band?
There was a problem hiding this comment.
OK, I think I understand, this would allow the band to be split up to four times. Useful for high frequency tones?
There was a problem hiding this comment.
Correct, the new bands are 16 and 24 wide instead of 18 and 22, so they can split much deeper.
phill-williams
left a comment
phill-williams
left a comment
There was a problem hiding this comment.
Reviewed, but I admit I don't really understand the changes in modes_fixed.h and modes_float.h, it would be helpful for someone with expertise in the codec to review.
Is there a unit test that validates the new functionality?
|
The static_modes_*.h files are auto-generated. The changes to those files are only a consequence of the eband5ms change in modes.c. |
|
Thanks. PR merged. |
2 participants
Removes the LN=-1 split limitation and changes the band layout so that all bands can now split all the way down to N=2 or N=3.