Changes in language UI and transcription languages. Bug fix #21
Conversation
Odrec
commented
Jan 28, 2025
•
Odrec
commented
- Adds German and English support to the interface
- Now the name of the language is used in the dropdown list for transcription languages instead of the abbreviation
- Adds all supported languages by whisperx that have alignment models as shown here: https://github.com/m-bain/whisperX/blob/main/whisperx/alignment.py
- Closes Translate the ui into german #14 and closes Write shortform for languages not in short form #15
- Fixes bug. Closes Higher minimum number of speakers than maximum #20
Add ffmpeg debian package to dockerfile
This patch adds a GitHub Actions workflow to automatically build container images and publish them to the GitHub container registry.
Automatically Publish Container Images
This patch sets sensible default values for some of the configuration options.
Configuration defaults
-Disable of streamlit usage statistics -Increase of max file size upload to 1GB -Hide developer options from hamburguer menu -Center media and editor on the screen -Only show transcription result message when there's an error -Show srt file as default -Added several help tooltips -Limited models to base and large-v3
Several interface and configuration changes.
Add logout button
…rent users using different files with the same name.
Use unique ids to save the output files to avoid conflicts with diffe…
Add README.md
Add missing screenshot
Changes in tooltips text to make them more on point.
Add logo .ico to browser tab
We are removing YouTube support for transcriptions due to known issues in the yt-dlp package with YouTube such as bot confirmations and ip blocks.
Delete YouTube
…ction's name switched to a string variable halfway through the code
…m number of speakers. Closes #20
Odrec
changed the title
|
@dennis531 I had to do some relatively significant changes to the code to keep the file from disappearing when changing language of the UI because of the limitations of the upload file widget from streamlit. I also realized there were some big files commited at some point so I deleted them which created the need to rebase the main branch and had to resolve some conflicts as a result. But I think now it should work. Can you test it and tell me what you think. |
|
Only German translations work, and only if I don't change the language: Screencast.from.2025-02-05.18-20-24.webm |
I realize I didn't upgrade the whisperx API server. The server was updated to accept the new list of languages so that's why it didn't work. It should be working now. |
|
Could you implement that the language does not change to Arabic when the UI language is changed? |
…the transcription variables.
…ge. Got rid of the widget form in the sidepanel to facilitate this behavior.
I had to change the widgets on the side panel to facilitate this behavior but it should be working fine now. Can you please test it when you have time? |
|
It works now but I found a new bug. The "Redo Transcription" action doesn't update the textbox with the transcriptions, see video. Screencast.from.2025-02-13.09-24-30.webm |
…e text editor to update with the new transcription. Got rid of some unnecessary session variables.
Ok, this should be fixed now. |
… ensure that the editor widget shows the new transcription.
…e cached values in the deployed instance that are different than in the local instance
…t updating its contents during deployment.
…on updating the quill editor. The time sleep was not ideal. Now the user can either reset the transcription without deleting the uploaded file or deleting the file which resets the entire transcription.
