Merge pull request #758 from drowe67/ms-linux-docs

Update README with Linux instructions for RADE.
pull/767/head
Mooneer Salem 2024-10-26 08:34:03 -07:00 committed by GitHub
commit 74f44f96a4
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 44 additions and 0 deletions

View File

@ -58,6 +58,50 @@ This document describes how to build the FreeDV GUI program for various operatin
$ ./build_linux/src/freedv
```
## Running RADE mode on Linux
RADE is a new FreeDV mode that uses machine learning to improve voice quality and decoding ability.
We are currently focused on Windows and macOS for initial development, but it is possible to run on
Linux by following these steps:
1. Install PyTorch, TorchAudio and matplotlib Python packages. Some distros have packages for one or more of these,
but you can also use pip in a Python virtual environment (recommended to ensure the latest versions):
```
$ cd freedv-gui
$ python3 -m venv rade-venv
$ . ./rade-venv/bin/activate
(rade-venv) $ pip3 install torch torchaudio --index-url https://download.pytorch.org/whl/cpu
(rade-venv) $ pip3 install matplotlib
```
*Note: you may need to install `python3-venv` or your distro's equivalent package in order to create Python virtual environments. Python 3.9+ is also required for PyTorch to work.*
2. Build FreeDV to make sure the correct dependencies are linked in (namely numpy):
```
(rade-venv) $ pwd
/home/<user>/freedv-gui
(rade-venv) $ ./build_linux.sh
```
3. Make sure FreeDV can find the ML model:
```
(rade-venv) $ pwd
/home/<user>/freedv-gui
(rade-venv) $ cd build_linux
(rade-venv) $ ln -s $(pwd)/rade_src/model19_check3 model19_check3
```
4. Execute FreeDV:
```
(rade-venv) $ pwd
/home/<user>/freedv-gui/build_linux
(rade-venv) $ PYTHONPATH="$(pwd)/rade_src:$PYTHONPATH" src/freedv
```
## Building without LPCNet
In preparation for possible future deprecation of FreeDV 2020 and 2020B modes, it is