search

bmcfee / resampy

Efficient sample rate conversion in python
https://resampy.readthedocs.io
ISC License
253 stars 36 forks source link

v0.3.0 produces other results than v0.2.2 #101

Closed hendriks73 closed 2 years ago

hendriks73 commented 2 years ago

We just stumbled over this by accident and it's (probably) completely inconsequential, but v0.2.2 produces slightly different results from v0.3.0.

To reproduce, unzip the archive and run:

conda env create -f resampy022.yml; conda env create -f resampy030.yml; conda activate resampy022; python resampy_test.py; conda activate resampy030; python resampy_test.py;

Perhaps this is an intentional side-effect of some other change.

Archive.zip

bmcfee commented 2 years ago

We just stumbled over this by accident and it's (probably) completely inconsequential, but v0.2.2 produces slightly different results from v0.3.0.

To reproduce, unzip the archive and run:

conda env create -f resampy022.yml; conda env create -f resampy030.yml; conda activate resampy022; python resampy_test.py; conda activate resampy030; python resampy_test.py;

Perhaps this is an intentional side-effect of some other change.

Archive.zip

Yes, this is known and expected. Per the changelog, the filters have been revised in the 0.3 release to reduce aliasing.

hendriks73 commented 2 years ago

Ah.. I guess that's:

Automate pre-computed filter generation. Regenerated and improved kaiser_fast and kaiser_best filters.

Cool. Thanks. Sorry to bother you with this.

bmcfee commented 2 years ago

No worries! Documentation probably could have been clearer here.