Bug 2573567 - Review Request: parrotia - Private, local audio and video transcription with Whisper
Summary: Review Request: parrotia - Private, local audio and video transcription with ...
Keywords:
Status: NEW
Alias: None
Product: Fedora
Classification: Fedora
Component: Package Review
Version: rawhide
Hardware: Unspecified
OS: Unspecified
unspecified
unspecified
Target Milestone: ---
Assignee: Nobody's working on this, feel free to take it
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard:
Depends On: 2573565 2573566
Blocks:
TreeView+ depends on / blocked
 
Reported: 2026-11-14 04:41 UTC by igor.rabbit99
Modified: 2026-11-16 15:06 UTC (History)
1 user (show)

Fixed In Version:
Clone Of:
Environment:
Last Closed:
Type: Bug
Embargoed:


Attachments (Terms of Use)

Description igor.rabbit99 2026-11-14 04:41:59 UTC
Spec URL: https://raw.githubusercontent.com/igorbispo99/ParrotIA/v1.4.0/packaging/fedora/parrotia.spec
SRPM URL: https://download.copr.fedorainfracloud.org/results/igorbispopessoal/parrotia/srpm-builds/11324219/parrotia-1.4.0-1.fc46.src.rpm

Description:
ParrotIA transcribes speech in audio and video files to text and subtitles
(TXT, Markdown, SRT, WebVTT, JSON) with Whisper models, running entirely on
the local computer. It provides a desktop application and a command-line tool
for batch jobs.

Speech models are downloaded on first use from Hugging Face, pinned to a
fixed revision and verified by SHA-256, into the user's data directory.

Fedora Account System Username: igorbispopessoal

COPR build (fedora-44, fedora-45, rawhide x86_64), including fedora-review output:
https://copr.fedorainfracloud.org/coprs/igorbispopessoal/parrotia/build/11324219/

Depends on the python-darkdetect (bug 2573565) and python-customtkinter
(bug 2573566) reviews.

I am the upstream author. Notes for the reviewer:
- Fedora does not ship faster-whisper, CTranslate2 or PyAV, so the package
  uses ParrotIA's whisper.cpp backend on python3-pywhispercpp; %prep drops
  those two requirements from pyproject.toml. ExclusiveArch follows
  python3-pywhispercpp, which is built for x86_64 only.
- Speech models are not packaged. The app downloads them on first use from
  Hugging Face, pinned to fixed revisions and checked against a SHA-256 table
  in the source; the AppStream description discloses this. The models are
  MIT-licensed (ggerganov/whisper.cpp, ggml-org/whisper-vad).
- Ships a desktop file, AppStream MetaInfo and icon (app ID
  io.github.igorbispo99.ParrotIA); %check runs desktop-file-validate,
  appstreamcli validate and the upstream test suite (mocked engines).
- rpmlint: 0 errors; warnings are no-manual-page-for-binary (parrotia,
  parrotia-gui) and python-leftover-require for python3-tkinter and
  python3-pillow-tk, which have no python3dist() provider.

Comment 1 igor.rabbit99 2026-11-16 15:06:45 UTC
Updated to ParrotIA 1.4.1. The only packaging-relevant change is that it now accepts CustomTkinter 6 (customtkinter >= 5.2.2, < 7), matching the python-customtkinter 6.0.0 update in bug 2573566. With the new python-customtkinter and python-darkdetect builds installed on Rawhide, the GUI starts and draws correctly under Xvfb.

Spec URL: https://raw.githubusercontent.com/igorbispo99/ParrotIA/v1.4.1/packaging/fedora/parrotia.spec
SRPM URL: https://download.copr.fedorainfracloud.org/results/igorbispopessoal/parrotia/srpm-builds/11335706/parrotia-1.4.1-1.fc46.src.rpm
Copr build (Fedora 44, 45 and Rawhide): https://copr.fedorainfracloud.org/coprs/build/11335706/


Note You need to log in before you can comment on or make changes to this bug.