|
Some checks failed
Linux Build / Linux cpu (push) Has been cancelled
Linux Build / Linux cpu-clang (push) Has been cancelled
Linux Build / Linux vulkan (push) Has been cancelled
mac-build / macOS CPU (push) Has been cancelled
Nix Build / Nix (cpu) (push) Has been cancelled
Nix Build / Nix (vulkan) (push) Has been cancelled
windows-build / Windows CPU (push) Has been cancelled
|
||
|---|---|---|
| .. | ||
| README.md | ||
| run_cpp.sh | ||
| run_server.sh | ||
| server.json | ||
| test_python.py | ||
OmniVoice from Python over HTTP
Drives OmniVoice from Python while inference stays in C++, by sending
OpenAI-compatible requests to audiocpp_server. The client uses only the Python
standard library.
Build
cmake -S . -B build -DENGINE_ENABLE_CUDA=ON
cmake --build build --parallel --target audiocpp_cli audiocpp_server
Direct C++ baseline
./tests/omnivoice/python_cpp_simple/run_cpp.sh
Writes tests/omnivoice/outputs/python_cpp_simple/cpp_output.wav.
Through the server
Terminal 1, and keep it open until the server reports it is ready:
./tests/omnivoice/python_cpp_simple/run_server.sh
Terminal 2:
python3 tests/omnivoice/python_cpp_simple/test_python.py
Writes tests/omnivoice/outputs/python_cpp_simple/python_output.wav.
Files
run_cpp.sh audiocpp_cli baseline
run_server.sh starts audiocpp_server with server.json
server.json server config, paths are relative to this file
test_python.py POST /v1/audio/speech client
Change weight types for the C++ baseline in run_cpp.sh, and for the server
path in the session_options block of server.json followed by a server
restart.