This directory contains input and configuration files for the ISAR selection tests based on S4-240396.
This directory contains input and configuration files for the ISAR selection tests based on [S4-240396](https://www.3gpp.org/ftp/TSG_SA/WG4_CODEC/TSGS4_127_Sophia-Antipolis/Docs/S4-240396.zip).
All tests use the BS.1534 (MUSHRA) test methodology.
@@ -56,3 +56,31 @@ Note: BS1534-2a requires FOA input files to generate MASA2DIR1. To use MASA2DIR2
- For head-tracking trajectories, the files must be named `item_name.wav.pre.csv` and `item_name.wav.post.csv` for split-prerendering and post-rendering trajectories respectively. The prerendering trajectory is only used for split rendering. The post rendering trajectory is used as the trajectory for the final stage, regardless of split rendering.
- For ISM metadata, files must be named `item_name.wav.0.csv` - `item_name.wav.3.csv` as for the regular IVAS selection tests.
3. Execute `python generate_test_isar.py BS1534-1a,a` where the input argument is the relevant experiment-lab pair. Currently only one lab per experiment is configured.
## Preprocessing for trajectory nullification
For setting up a test with trajectory nullification, the input items must be preprocessed according to the specific trajectory.
This preprocessing consists of two main steps. First the trajectory must be inverted to apply the inverse rotation to the source audio.
This can be done with the help of the [trajectories submodule](../../ivas_processing_scripts/trajectories/README.md) by running the command:
Then the input item needs to be rotated in place using the inverted trajectory, which can be performed using the `--in_trajectory` argument of the [audiotools module](../../examples/audiotools.ipynb):
Inversion may be performed directly for SBA and MC formats.
For ISM format it is only possible to perform this on a 20ms basis due to metadata granularity.
For MASA format, the masaAnalyzer must be run on an inverted SBA input to generate the corresponding MASA transport channels and metadata.
Refer to [_Tdoc S4-240254: Trajectory Nullification for Binaural Renderer Evaluation, Fraunhofer IIS_](https://www.3gpp.org/ftp/TSG_SA/WG4_CODEC/TSGS4_127_Sophia-Antipolis/Docs/S4-240254.zip) for further details.
After generation of the "inverted" item, this may be directly used as input for the processing scripts.