ips-lmu/transcription-portal

A portal that offers a transcription chain for multi upload and processing of audio files using ASR, OCTRA, MAUS and EMU-webApp.

TranscriptionPortal

Implementation of a transcription chain that supports ASR, OCTRA, MAUS and EMU-webApp.

Website

You can use TranscriptionPortal without installation here: https://www.phonetik.uni-muenchen.de/apps/TranscriptionPortal/

Installation

You can install TranscriptionPortal on your own web server. Just download the repository and copy the contents from the dist folder to your web server.

Development

Installation

Clone this repository and install all dependencies using npm install. It is recommended to use WebStorm or PHPStorm IDE and the tslint settings from tslint.json in order to check the code style automatically.

Contribution

For this project commitizen is used. If you want to contribute to this project you should make use of this tool to create commit messages important for the changelog (other commit messages will be ignored otherwise). For WebStorm or PHPStorm there is a Commit Template Plugin to create these templates.

For creating changelog automatically, conventional-changelog-cli is used. To create changelogs you need to install conventional-changelog-cli globally as described on its github page. After installation you just need to run npm run changelog to create the changelog automatically.

Development server

Run ng serve for a dev server. Navigate to http://localhost:4200/. The app will automatically reload if you change any of the source files.

Build

Run sh ./build.sh to build the project. The build artifacts will be stored in the dist/transcription-portal directory. Adapt the configuration in the build.sh script as you need.

transcription-portal's Issues

archive modal: add support for conversions

For now there is no option to select conversions of the results. A list of options is needed here.

transcripts: saving transcripts to IndexedDB

IndexedDB could save data up to 2GB. That enables to save the transcript files to IndexedDB in order to keep transcripts even if they are not on the server currently.

• Save the files's content (excl. audio files) after DragnDrop to the IDB
• Download and save transcript to IDB after operation finished
• Download & save transcript files which are not already saved to IDB
• Use BlobURL instead of web URL for results
• Show red unlink item only if transcript or audio file does not exist on the server AND in the IDB
• Reupload for audio file
• Before operation started: check if transcript is online otherwise reupload before processing
• Archive feature must reupload local transcripts for zAPI

This improves usability and enables downloading the results long after 24h.

show loading spinner on startup

appendings: remove selected appendings

There should be a way to remove added appendings like AnnotJSON. Perhaps something like an entry in the contextmenu is needed.

add supported audio duration, file size etc.

It would be good for users to get information about each service provider:

• max. signal length
• max file size

set shortcuts for windows and mac

• select tasks while pressing CTRL or CMD only
• set standard shortcuts for removing, selecting more items at once

add option for renaming zip archives

Before a download has started there should be an option to rename the zip archive.

notifications: show finished file name in message

keep sort order after delete a task from folder

When a task is deleted from a TaskDirectory the remaining task should have the same list position as before.

Because this needs a sorting attribute it's also related to #24.

appendings: overwrite old appendings

When the user adds a new appending and another appending already exists, the old one should be replaced.

processing can be done without wav file

keep original file name for download

It's better for the user if the downloaded file name is the same as the original audio file. The user should'nt see the internal used escaped file name.

convert date and time strings according to browser language and timezone

feature: start/stop processing via button

combining audio file + annotJSON does not work

queue modal: show appendings

queue-modal: feature for setting sort order

design: queue modal does not shown correctly on resizing

work-flow: options to disable MAUS and EmuWebApp

Sometimes a user like to use the steps up to OCTRA only. In this situation, MAUS and EMU-webApp should be disabled.

download result: open in new tab

OCTRA doesn't load on first try. No chance to reload OCTRA

make sidebar resizeable

sort order of tasks

• add sortorder attribute to IDB
• unsorted is -1
• the sortorder represent the index position in the list of tasks
• to avoid calculation overhead, the sortorder is saved whenever a task or directory is added/inserted
• change sort order via context menu on the proceedings table
• change sort order via arrows in the queue modal

save default sidebar width set by the user

This could be saved automatically to the IDB and improves usability

show alert if authentication is expired

resume does not reset processing duration

archive download: make archive of selected lines

1. User select lines
2. Right click > context menu > click on "Download all results"
3. Archive modal is shown
4. Selection for further conversions
5. Button to remove selected lines after download

The best way is to implement two modes: one for column selection and one for line selection.

close and open folders

creating archives not working properly

If the user creates an archive and after that he changes the options for the archive, the download link is still the old one.

add option to retry a failed operation

preview window for transcripts

add alert if permission for push notification was denied

new versions of Chrome and Firefox won't show any popup if push notifications was denied. That's why in this case an alert should appear to inform the user.

handling folders doesn't work in some situations

use Dexie.js instead of current implementation

https://dexie.org/

octra: first interaction problem

Reproduction

1. User edit a new file with OCTRA.
2. User does not click on the "Save transcription" button and clicks on the back button
3. User reloads the OH-Portal
4. The OCTRA column is empty and the user can't continue with OCTRA

Expected

The user must have the option to continue the aborted OCTRA operation after reload.

results table broken if files do not exist on the server

insert revokeObjectURL() calls after URLs are not needed anymore

create config.json for individual settings

filenames in folders not cropped on resize

improve visible error messages

• simplify error messages
• show different colors for each type (warning, error etc.)

popover is shown on tool loading

the popover should be hidden when a tool loads.

design: animation when a tool shows up

double extensions in file names

combination of audio file + transcript doesn't work

context menu: change label for selected items

reupload for non existing files on the server

add browser check

Add a browser check to prevent Safari browsers because it's not compatible

upload: more than one uploads concurrently allowed

results table: sort results descending

this improve usabilty because the latest result is shown at the top

tool-loader opens accidentally without user's action