Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/schnoddelbotz/whisper-ui
Transcribe audio/video to text, locally on macOS, Linux and Windows. A simple whisper.cpp wrapper/UI built with Go/Fyne.
https://github.com/schnoddelbotz/whisper-ui
ffmpeg ffmpeg-wrapper fyne gui local privacy speech-to-text transcription whisper whisper-cpp
Last synced: 1 day ago
JSON representation
Transcribe audio/video to text, locally on macOS, Linux and Windows. A simple whisper.cpp wrapper/UI built with Go/Fyne.
- Host: GitHub
- URL: https://github.com/schnoddelbotz/whisper-ui
- Owner: schnoddelbotz
- Created: 2024-11-19T10:54:34.000Z (about 1 month ago)
- Default Branch: main
- Last Pushed: 2024-12-08T16:21:00.000Z (15 days ago)
- Last Synced: 2024-12-08T17:24:38.779Z (15 days ago)
- Topics: ffmpeg, ffmpeg-wrapper, fyne, gui, local, privacy, speech-to-text, transcription, whisper, whisper-cpp
- Language: Go
- Homepage:
- Size: 1.78 MB
- Stars: 1
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# whisper-ui
whisper-ui is a very simple [whisper-cpp](https://github.com/ggerganov/whisper.cpp) GUI wrapper.
It enables local/offline usage of [Whisper](https://openai.com/index/whisper/) to transcribe audio
or video input files to plain text. whisper-ui uses [Fyne](https://fyne.io/) to build the UI.
[Download a whisper-ui release](https://github.com/schnoddelbotz/whisper-ui/releases),
available for macOS, Windows and Ubuntu. Releases bundle a whisper-cpp and
[ffmpeg](https://www.ffmpeg.org/download.html) executable to free users from
any further setup/compilation tasks.
[Models for whisper-cpp](https://github.com/ggerganov/whisper.cpp/blob/master/models/README.md)
can be downloaded using the whisper-ui app.
## Notes - macOS
Note that the releases built via github [workflow](.github/workflows/release.yaml) are not signed.
For macOS, this means you have to remove quarantine flag (using `xattr -d com.apple.quarantine ...`).
To build whisper-ui from source:
```bash
git clone https://github.com/schnoddelbotz/whisper-ui.git
cd whisper-ui
make build-darwin
```
The build / Makefile fetches a [static ffmpeg build](https://evermeet.cx/ffmpeg/),
builds whisper-cpp and puts both into the .app bundle.
## Notes - Linux
The Linux release contains a Makefile. It can be used to install or uninstall
whisper-ui as Desktop app. For a system-wide install, use `make install`.
For an installation for the current user, use `make user-install`.
The Linux build expects ffmpeg to be installed via package manger / in `$PATH`.
If not already present, use `sudo apt install ffmpeg`.
To build whisper-ui from source:
```bash
git clone https://github.com/schnoddelbotz/whisper-ui.git
cd whisper-ui
make build zip-linux
```
## Notes - Windows
Like the macOS release, whisper-ui.exe is not signed - accordingly
Defender will report a "unrecognized app" and clicking "More info"
will be required while running it for the first time.
For building whisper-ui from source, see the
[release.yaml](.github/workflows/release.yaml)'s windows section.
## todo / issues
Status: Works for me (Sonoma/ARM, Monterey/x86_64), will possibly look into improvements.
- could skip ffmpeg conversion if input is already a 16kHz WAV
- more verbose output/progress feedback from ffmpeg and whisper-cpp?
- Sequioa/ARM using x86 build gives `Bad CPU Type in Executable`? `softwareupdate --install-rosetta`.
- option to "open output upon completion" may not work on Windows, same for Markdown link to file
- it would be nice to be able to [select multiple files](https://github.com/fyne-io/fyne/issues/1082) for conversion
- ahem, tests?
## license
MIT