Created attachment 177148 [details] Displays what I describe as a "persistent stop" window. SUMMARY Cant install Whisper turbo models on Fedora41, When I click 安装一个语音模型 - 我们推荐使用 turbo ——> "下载（Download 1.4GB）" Then：An new Windows show This”正在安装语音模型 turbo“(Installing speech model turbo) but everyting stop on this Windows. My proxy software (clash) can shows the current network traffic status, so I'm sure there's no actual download request happening. There are also no errors reported, he just stops. STEPS TO REPRODUCE 1. install kdenlive last (24.12.0.-1.fc41) 2. conda create -n kdenlive-whisper python=3.12 conda activate kdenlive-whisper pip install -r /usr/share/kdenlive/scripts/whisper/requirements-whisper.txt 3. launch by : bash -c "source ~/miniconda3/bin/activate kdenlive-whisper && kdenlive %F" 4. to setting find speech_to_text OBSERVED RESULT EXPECTED RESULT The model file should be downloaded and installed, or stopped on any error and an error message reported to the user. SOFTWARE/OS VERSIONS (kdenlive-whisper) [xfox@JiaoLong5-76s ~]$ kinfo Operating System: Fedora Linux 41 KDE Plasma Version: 6.2.5 KDE Frameworks Version: 6.9.0 Qt Version: 6.8.1 Kernel Version: 6.12.7-200.fc41.x86_64 (64-bit) Graphics Platform: Wayland Processors: 16 × AMD Ryzen 7 5800H with Radeon Graphics Memory: 31.2 GiB of RAM Graphics Processor: NVIDIA GeForce RTX 3060 Laptop GPU/PCIe/SSE2 ADDITIONAL INFORMATION If you need to add any information, I will be happy to provide it as soon as possible. Usually I will return to this site every day. If you want to find me faster, you can leave a message on my blog (xfox.fun) and I will receive a text message immediately.
All of Whisper models cant setup ,not only turbo.
Thank you for reporting. According to the screenshot only a quarter of the turbo model is downloaded. You have to be patient it may take 30 minutes or more to download depends on your internet connection and server availability. If it still not works try with the daily AppImage (we have made some Whisper updates) https://cdn.kde.org/ci-builds/multimedia/kdenlive/master/linux/
(In reply to emohr from comment #2) > Thank you for reporting. According to the screenshot only a quarter of the > turbo model is downloaded. You have to be patient it may take 30 minutes or > more to download depends on your internet connection and server availability. > > If it still not works try with the daily AppImage (we have made some Whisper > updates) https://cdn.kde.org/ci-builds/multimedia/kdenlive/master/linux/ I'll try to let him continue downloading overnight, which should ensure he has enough time to finish. If the progress remains there, I'll report back.(In reply to emohr from comment #2) > Thank you for reporting. According to the screenshot only a quarter of the > turbo model is downloaded. You have to be patient it may take 30 minutes or > more to download depends on your internet connection and server availability. > > If it still not works try with the daily AppImage (we have made some Whisper > updates) https://cdn.kde.org/ci-builds/multimedia/kdenlive/master/linux/ (In reply to emohr from comment #2) > Thank you for reporting. According to the screenshot only a quarter of the > turbo model is downloaded. You have to be patient it may take 30 minutes or > more to download depends on your internet connection and server availability. > > If it still not works try with the daily AppImage (we have made some Whisper > updates) https://cdn.kde.org/ci-builds/multimedia/kdenlive/master/linux/ I'll try to let him continue downloading overnight, which should ensure he has enough time to finish. If the progress remains there, I'll report back.
(In reply to emohr from comment #2) > Thank you for reporting. According to the screenshot only a quarter of the > turbo model is downloaded. You have to be patient it may take 30 minutes or > more to download depends on your internet connection and server availability. > > If it still not works try with the daily AppImage (we have made some Whisper > updates) https://cdn.kde.org/ci-builds/multimedia/kdenlive/master/linux/ kdenlive-master-9479-linux-gcc-x86_64.appimage numba==0.60.0 doesn't support Python 3.13, so I had to use a virtual environment, but he couldn't call the Python provided by the virtual environment correctly.
Deleting and re-creating a new virtual environment is effective (but its Chinese translation looks a bit lame compared to the version I use. This may be due to some settings of the Python virtual environment, such as deletion and reconstruction, which are included in the "plugin" and cause misunderstandings. In short, the logic is not too strange for Chinese users). I am sure that the current latest version has solved the problem I described. Thank you very much for the help provided by community members, and I hope Fedora will push the latest version to the warehouse as soon as possible.
Another unexpected surprise is that after completing the speech-to-text configuration of the Python virtual environment created by the current latest version (kdenlive-master-9479-linux-gcc-x86_64.appimage), the old version packaged in Fedora that I am currently using can also use the related functions normally by specifying this configured new virtual environment.