Skip to content

v0.1.0-alpha.1 (アルファ版)

Pre-release
Pre-release

Choose a tag to compare

@taoman26 taoman26 released this 22 Oct 06:23
· 35 commits to main since this release
56ed70d

AppImage (Linux, x86-64)

AudioKnowledgeExplorer-<version>-x86_64.AppImage をダウンロードして、実行権限を付けてダブルクリックするだけで動きます(インストール不要)。

chmod +x AudioKnowledgeExplorer-*.AppImage && ./AudioKnowledgeExplorer-*.AppImage
  • 動作条件: x86-64 の Linux、glibc 2.35 以上(Ubuntu 22.04 以降が目安)、AVX2 / FMA / F16C 対応の CPU(2013 年以降)。FUSE 2(Ubuntu では libfuse2)が必要です。無いときは --appimage-extract-and-run を付けて実行してください。
  • 表示は X11(xcb)です。Wayland のデスクトップでは XWayland を使います。
  • whisper のモデルは同梱していません。 初回起動時に、置き場所と入手先が画面に出ます(README)。日本語には ggml-large-v3-turbo.bin をお勧めします。
  • 日本語入力(fcitx5)用のプラグインが入っています。

ライセンスとソース

アプリ本体は MIT です。この AppImage は、GPL の部品を含みます(Ubuntu 22.04 の FFmpeg は GPL v2+ のビルド、GStreamer の plugins-ugly など)。同梱しているライブラリと、元の Ubuntu のパッケージの一覧は、AppImage の中の usr/share/doc/AudioKnowledgeExplorer/BUNDLED-PACKAGES.txt にあり、ソースは apt source <パッケージ名>(Ubuntu 22.04)で入手できます。アプリ本体のソースは、このタグの Source code(zip / tar.gz)です。ビルド手順は packaging/appimage/ にあります。

既知の制限

アルファ版です。文字起こしの精度はモデルに依存します。詳しくは README の「既知の制限」を見てください。


What's Changed

  • Fix "(音声)"-only transcripts, add homophone search, folder removal and watch setting by @taoman26 in #1
  • README: build MeCab and the JUMAN dictionary from source on Haiku by @taoman26 in #2
  • Default the transcription language to Japanese and advise on the model by @taoman26 in #3
  • README: Japanese input on Ubuntu 22.04 with fcitx5 (no Qt6 plugin) by @taoman26 in #4
  • AppImage for Linux, FFmpeg 4.4 support, release workflow by @taoman26 in #5

New Contributors

Full Changelog: https://github.com/taoman26/AudioKnowledgeExplorer/commits/v0.1.0-alpha.1