xref: /aosp_15_r20/external/libopus/dnn/torch/rdovae/README.md (revision a58d3d2adb790c104798cd88c8a3aff4fa8b82cc)
1# Rate-Distortion-Optimized Variational Auto-Encoder
2
3## Setup
4The python code requires python >= 3.6 and has been tested with python 3.6 and python 3.10. To install requirements run
5```
6python -m pip install -r requirements.txt
7```
8
9## Training
10To generate training data use dump date from the main LPCNet repo
11```
12./dump_data -train 16khz_speech_input.s16 features.f32 data.s16
13```
14
15To train the model, simply run
16```
17python train_rdovae.py features.f32 output_folder
18```
19
20To train on CUDA device add `--cuda-visible-devices idx`.
21
22
23## ToDo
24- Upload checkpoints and add URLs
25