A fast, partitioned convolution engine library
Convolution engine based on FFT convolution and using non-uniform partition sizes: small ones at the start of the IR and building up to the most efficient size further on. It can perform zero-delay processing with moderate CPU load.
Main features:
* Any matrix of convolutions between up to up 64 inputs and 64 outputs, as long as your CPU(s) can handle it.
* Allows trading off CPU load to processing delay, and remains efficient even when configured for zero delay.
* Sparse and diagonal matrices are handled as efficiently as dense ones. No CPU cycles or memory resources are wasted on empty cells in the matrix, nor on empty partitions if IRs are of different length.
- Developed at multimedia:libs
- Sources inherited from project openSUSE:Factory
-
1
derived packages
- Download package
-
Checkout Package
osc -A https://api.opensuse.org checkout openSUSE:Backports:SLE-15-SP4:FactoryCandidates/zita-convolver && cd $_ - Create Badge
Refresh
Refresh
Source Files
| Filename | Size | Changed |
|---|---|---|
| baselibs.conf | 0000000019 19 Bytes | |
| zita-convolver-4.0.0.tar.bz2 | 0000020268 19.8 KB | |
| zita-convolver.changes | 0000001570 1.53 KB | |
| zita-convolver.spec | 0000003566 3.48 KB |
Revision 2 (latest revision is 3)
Yuchen Lin (maxlin_factory)
accepted
request 616633
from
Mia Herkt (lachs0r)
(revision 2)
- Add baselibs.conf
- Update to version 4.0.0
1. Version 4 now uses int types from <stdint.h> internally and for
the arguments of all member funcions.
2. The set_density() function has been removed, and the matrix
density hint is now an additional argument to configure().
If your application does not use impdata_update(), then (1)
and (2) are the only relevant changes.
3. The way impdata_update() works has changed.
This function is used to modify IR data while the convolver is
actually running. It does not use any memory allocation nor
modify internal data structures, and only data in already
existing partitions can be modified this way.
In versions <= 3, this function would *overwrite* any existing
data. From version 4, impdata_update() *adds* to existing IR
data, just as impdata_create() does. So in order to replace an
existing IR, you first need to clear it using the new function
impdata_clear(). This will clear (but not delete) all IR data
for a given input, output pair.
- Replace %soname by %sover to better reflect its use.
- Drop bias from description.
- Add Group: line for shared library subpackage.
Comments 0