Arch Linux User Repository

@jatinkrmalik apologies I did not mean to imply that you maintained the python-whispercpp packages, or that the issue with them were attributed to you in any way, but only to point out that the issue I was having was related to them. I have reached out to goodroot and it looks like those packages are out dated an may not be updated, but that is up to goodroot. I have since switched to your install script which does a better job of building this locally rather than using the AUR package.

tnx. solved.

@simona @Masalababa @avocadoboat @galvez_65: 0.16.1 is tagged and I just pushed it to the AUR git repo.

yay -Syu vocalinux

@simona: ERROR Missing dependencies: setuptools<82,>=61 was python -m build --no-isolation against Arch extra python-setuptools 84. The <82 cap is gone.

@avocadoboat @Masalababa: the whisper_full_params / context_params crash is AUR python-pywhispercpp-cpu/cuda/rocm still being 1.4.x. Those wheels accept the kwarg in Python and then setattr it onto the native object, which raises. 0.16.1 skips that argument on pywhispercpp older than 1.5 and still loads (default GPU). Same as GitHub #625.

@galvez_65: there is still no AUR python-pywhispercpp-vulkan. The backends this package depends on are 1.4.x. Vulkan 1.5 is install.sh / AppImage, not this AUR package. If you want those PKGBUILDs upstreamed, that belongs on the python-pywhispercpp AUR packages (maintainer: goodroot), not here.

Notes: https://github.com/VocaHQ/vocalinux/releases/tag/v0.16.1

ERROR Missing dependencies: setuptools<82,>=61

I got the same error on Manjaro as described by avocadoboat. Starting Vocalinux from terminal, here is the output:

2026-08-19 21:36:48,940 - vocalinux.single_instance - INFO - Single instance lock acquired

2026-08-19 21:36:52,427 - vocalinux.utils.resource_manager - INFO - Found resources directory: /usr/lib/python3.14/site-packages/vocalinux/resources

2026-08-19 21:36:54,347 - vocalinux.ui.logging_manager - INFO - Logging manager initialized

2026-08-19 21:36:54,347 - vocalinux.ui.logging_manager - INFO - Global logging manager initialized

2026-08-19 21:36:54,348 - vocalinux.main - INFO - Logging system initialized

2026-08-19 21:36:54,379 - vocalinux.ui.config_manager - INFO - Loaded configuration from /home/space/.config/vocalinux/config.json

2026-08-19 21:36:54,379 - vocalinux.main - INFO - Using engine=whisper_cpp (from saved config)

2026-08-19 21:36:54,379 - vocalinux.main - INFO - Using language=auto (from saved config)

2026-08-19 21:36:54,379 - vocalinux.main - INFO - Using model=base (from saved config)

2026-08-19 21:36:54,379 - vocalinux.main - INFO - Final settings: engine=whisper_cpp, language=auto, model=base

2026-08-19 21:36:54,379 - vocalinux.main - INFO - Initializing Vocalinux...

2026-08-19 21:36:56,310 - vocalinux.speech_recognition.silero_vad - INFO - Silero VAD model loaded

2026-08-19 21:36:56,310 - vocalinux.speech_recognition.recognition_manager - INFO - Using Silero neural VAD

2026-08-19 21:36:56,310 - vocalinux.speech_recognition.recognition_manager - INFO - Initializing speech recognition with whisper_cpp engine, auto language and base model

2026-08-19 21:36:57,084 - vocalinux.utils.whispercpp_model_info - INFO - Vulkan support detected: Intel(R) HD Graphics 4600 (HSW GT2)

2026-08-19 21:36:57,084 - vocalinux.speech_recognition.recognition_manager - INFO - whisper.cpp backend selection priority: Vulkan -> CUDA -> CPU

2026-08-19 21:36:57,084 - vocalinux.speech_recognition.recognition_manager - INFO - whisper.cpp using Vulkan GPU backend: Intel(R) HD Graphics 4600 (HSW GT2)

2026-08-19 21:36:57,084 - vocalinux.speech_recognition.recognition_manager - INFO - Using Vulkan GPU [0]: Intel(R) HD Graphics 4600 (HSW GT2)

2026-08-19 21:36:57,173 - vocalinux.speech_recognition.recognition_manager - INFO - whisper.cpp hardware: vulkan | Intel(R) HD Graphics 4600 (HSW GT2) | RAM: 3GB

2026-08-19 21:36:57,173 - vocalinux.speech_recognition.recognition_manager - INFO - whisper.cpp model file: /home/user/.local/share/vocalinux/models/whispercpp/ggml-base.bin (141.1 MB)

2026-08-19 21:36:57,173 - vocalinux.speech_recognition.recognition_manager - INFO - Loading whisper.cpp 'base' model... 2026-08-19 21:36:57,176 - vocalinux.speech_recognition.recognition_manager - WARNING - pywhispercpp does not support 'no_timestamps'; removing from model kwargs. 2026-08-19 21:36:57,176 - vocalinux.main - ERROR - Failed to initialize Vocalinux: '_pywhispercpp.whisper_full_params' object has no attribute 'context_params' and no dict for setting new attributes

2026-08-19 21:36:57,176 - vocalinux.main - ERROR - Please check the logs above for more details

Finally got Vocalinux 2 install, The issue wasn't Vocalinux, but rather python-pywhispercpp. I ended up having to create a PKGBUILD for python-pywhispercpp-vulkan against the current 1.5.0 version, and in the process I had to create a couple more custom PKGBUILD files for some missing python packages. Happy to share if anyone is interested. Note I'm sure the PKGBUILD files can be improved.

My installation succeeds, but when I try to start vocalinux is fails with the following error, which seems to indicate some form of dependency hell:

vocalinux.main - ERROR - Failed to initialize Vocalinux: '_pywhispercpp.whisper_full_params' object has no attribute 'context_params' and no __dict__ for setting new attributes

@XSilverTH: Thanks for the report.

vocalinux depends on the virtual package python-pywhispercpp instead of a specific backend, so you can swap the CPU, CUDA, or ROCm build without pacman treating it as a broken dependency. That change is documented in GitHub issue 578.

On a fresh install, yay/paru will ask which provider to use:

1. python-pywhispercpp-cpu (default)
2. python-pywhispercpp-cuda
3. python-pywhispercpp-rocm

Accept the default unless you have an NVIDIA GPU and want CUDA.

Option 3 is what failed here. python-pywhispercpp-rocm still requires rocm-hip-runtime<7 at runtime, and rocm-hip-sdk<7 plus rocm-opencl-sdk<7 to build. Those names are in extra, not the AUR. extra currently has rocm-hip-sdk 7.2.4-1, which does not satisfy <7, and there is no AUR package that does. The helper then reports that it cannot find rocm-hip-sdk<7.

That is a bug in python-pywhispercpp-rocm (maintainer: goodroot), not in vocalinux. The version pin is there because ggml's HIP backend still does not compile against ROCm 7; see whisper.cpp issue 3553. Until that package is rebuilt for current ROCm, AMD GPU acceleration is not available through this AUR path. Use option 1.

:: There are 3 providers available for python-pywhispercpp: :: Repository AUR 1) python-pywhispercpp-cpu 2) python-pywhispercpp-cuda 3) python-pywhispercpp-rocm

Enter a number (default=1): ==> 3 -> No AUR package found for rocm-hip-sdk<7 (required by: python-pywhispercpp-rocm) -> No AUR package found for rocm-opencl-sdk<7 (required by: python-pywhispercpp-rocm) -> No AUR package found for rocm-hip-runtime<7 (required by: python-pywhispercpp-rocm) -> could not find all required packages: rocm-hip-sdk <7 could not find all required packages: rocm-opencl-sdk <7 could not find all required packages: rocm-hip-runtime <7

Text injection on X11 is not working for me with this release. The AppImage of 0.15.0 does work, any chance of getting an AppImage based package?