ParrotIA

Transcribe audio and video recordings on your computer with ParrotIA. This initial edge preview is version 1.3.1. Install it with:

 sudo snap install parrotia --edge

Use the desktop interface for individual files or folders, or the command line for scripted jobs. Export plain text, Markdown, SRT subtitles, WebVTT subtitles and JSON. Detect the spoken language automatically or choose it, compare model speeds, and cancel running jobs.

Transcription runs locally. Your audio is not uploaded to a transcription service. No account or API key is needed to use the app. The first use of each Whisper model downloads it from Hugging Face and its storage services; cached models can then be used offline. Downloads range from about 75 MB to 3.1 GB. The default large-v3-turbo model is about 1.6 GB; allow extra space for the download and cache.

This Snap supports CPU inference on 64-bit Intel/AMD Linux desktops. The GUI uses X11/XWayland. NVIDIA acceleration is not part of its supported configuration.

Launch ParrotIA from the application menu or run parrotia.gui. For command-line help, run parrotia --help. Models are stored under ~/snap/parrotia/common/huggingface/ and survive refresh/revert.

Non-hidden files you own in your home folder are accessible. Mounted external drives require the optional removable-media connection:

 sudo snap connect parrotia:removable-media

Choose a writable output folder when the source directory is read-only.

ParrotIA is free software under AGPL-3.0-or-later. It uses shared FFmpeg libraries under LGPL-2.1-or-later. The exact native source archives, build settings and notices are included at /snap/parrotia/current/share/doc/parrotia/native/; application source and additional notices are under /snap/parrotia/current/share/doc/parrotia/ and /snap/parrotia/current/share/licenses/parrotia/. Third-party components retain their own licenses.

Source code and support: https://github.com/igorbispo99/ParrotIA