Fennec 0.3.0

Install on Ubuntu 24.04 or newer (or Debian with GTK 4.14+ and libadwaita 1.5+):

sudo apt install ./fennec_*.deb

The speech model runs on the GPU through Vulkan (AMD, Intel and NVIDIA) when there is one, and on the CPU otherwise. Settings → Speech model → Compute shows which, and can force the CPU.

Fennec Recorder (fennec-recorder-*.apk, Android 10 or newer) records on your phone and sends the audio to Fennec over your own network. Install the APK, then in Fennec choose Settings → Phone → Pair a phone and scan the code.

What's Changed

• Level meter: speech rises and falls instead of filling every bar by @AngelFreak in #7
• Dictate and transcribe in English as well as Danish by @AngelFreak in #8
• Build whisper.cpp with its llamafile matrix kernels (1.4× faster on CPU) by @AngelFreak in #9
• Download Edda as q4_k: as accurate, 100 MB smaller, 1.5× faster by @AngelFreak in #10
• fennec-bench: time --fast and --prompt as scored by @AngelFreak in #11
• Version 0.3.0 by @AngelFreak in #12

Full Changelog: v0.2.1...v0.3.0