Getting started

Preparation

The transcriber makes use of the Alveo Virtual Laboratory to provide a library of speech and language data. If you haven't already, you should register an account.

Registering an account allows you to:

• authenticate the transcriber application
• create item lists to work with
• work with item lists shared by other users
• synchronise your data
• allow you to revert changes & explore change history
• utilise upcoming features such as auto-transcription, transcription collaboration

When your account is ready for use, you will be able to access the transcriber.

Authentication

Point your browser to https://alveo.github.io/alveo-transcriber/. The transcriber uses OAuthenciation to securely connect with the Alveo website, so you will find that once on the transcriber page you will be requested to log in, and you will be redirected to the Alveo website to do so.

If it is your first time using the application, you will be prompted to allow the application to access non-sensitive user data.

Note: The Alveo Transcriber makes use of Alveo Transcriber Services (ATS) which is a detached environment from the Alveo Virtual Laboratory, designed to provide extra functionality such as synchronisation and auto-segmentation. In order to authenticate you successfully across these services, your API key will be securely sent upon each request, but never stored. ATS does not store or collect sensitive data and the only information it tracks about a user is their alveo user ID. For more information, please feel free to browse the alveo-transcriber-services repository.

Accessing data

After signing in, you should be prompted to Download Alveo data. Please touch/click that button to continue.

When the download is complete, you should have a view of all of your item lists, as well as ones shared by other users. The Alveo service makes use of these itemlists to work with data. Currently, the transcriber does not support a means of working with items that are outside of a list. To create a list, you will have to head to the main Alveo website (https://app.alveo.edu.au/) and proceed to one of the collections.

Caching

The transcriber works as offline-first application. This means the transcriber will cache the data that gets downloaded as you use the application. The main advantage of this means that you can take the transcriber with you where you go even if you do not have reliable (or any) internet access, assuming you have downloaded the data in advance.

Once you have data to work with, you won't be prompted to log in unless you attempt to download some data which has not yet been cached. The application will prompt you to optionally log in to download data, otherwise you can safely click off the dialogue and continue with another activity.

Ideally, you should remain logged in at all times to ensure that your changes get synchronised. This will mean that they're available across all devices. It will also give you access to features such as autosegmentation and revisioning.

Preparing to transcribe

Start by selecting an item list then finding an item you wish to subscribe. The application supports searching by Alveo identifier/name and is important to use when working with large amounts of data.

Once the data has been downloaded, you will note that there is an audio source dropdown available. By default, the most preferred audio source has been selected for you. If you need to change the audio source, do so in this area. Changing the audio source will not reset your transcription and you will be able to work with the same transcription regardless of which audio source you have selected.

Once ready, click the Transcribe button.

Note: Only the selected audio source will ever be downloaded. This is important to know if you need to cache multiple audio sources.

Transcribing

Layout

The transcribing screen will present to you the following:

• Revisions (explained further down later)
• Time of last save (which will make note if the save has not been synchronised (likely because you are not logged in) but has been saved locally).
• The identifier of the chosen item.
• The toolbar, containing controls to: autosegment, export, exit.
• A waveform. This is where you can create, resize and edit regions by dragging your mouse (or finger for touch supported devices) along the waveform.
• Player controls: pausing, playing, zoom and repeat last X seconds.
• The autoplay checkbox. This will automatically play the audio of the caption, each time you select one.
• Editor toggle. This will switch between the experimental 'mobile-friendly' caption editor and the standard list view. The 'mobile-friendly' caption editor while functional, is not actively being worked on. We'd love to hear your thoughts on whether such a thing is useful to you.
• The transcription area. This is where when you have some captions created on the waveform, you will see areas to fill in text, play or delete segments.

Hotkeys

The transcriber makes use of Tab and Shift-Tab to go to the next and previous segment respectively. The Escape key can be used to replay the current segment. The Delete key provides a shortcut to prompt deletion of a caption.

Autosegment

The transcriber makes use of the Alveo Transcriber Services (ATS) to automatically segment audio files for your convenience. Use of this service requires you to be logged in and with an active internet connection. To use the autosegmenter, click the Autosegment button at the top of the transcriber, in the controls area. Please take important note of the dialogue that warns you that any existing transcriptions on the item you are transcribing, will be replaced by what the service produces. If you have made a mistake or do not like the output, you can revert your changes (see below: 'Revisions' section).

Saving & Exporting

Data including transcriptions will get saved locally inside of your webbrowser offline cache. If you are logged in, it will also be synchronised to the Alveo Transcriber Services (ATS). Synchronised data can be accessed across multiple devices. Transcriptions can be exported from the cache via the Export CSV and Export JSON buttons. Importing transcriptions is currently not possible. Being able to publish transcriptions is a planned and upcoming feature.

Important: Clearing browser history may destroy your transcriptions. Please make sure they are synchronised with ATS before doing so.

Mobile view

There is a basic mobile transcriber view available. The button is located next to the autoplay checkbox, at the top right of the transcription area. You can return to the list view by pushing the same button again.

The mobile view is designed to display more effectively on screens of all shapes and sizes, however it is mainly intended to be used on devices without a large screen. Swipe support to move between annotations is planned.

Revisions

If you have been working on a transcription while logged in, you will have access to your revision history for that transcription. In the top right corner, you will note 'Current revision'. Click/press Change to bring up the Revision Selector. Here, you will see your revision history. Scroll down to a more relevant revision and then click/press it. After, hit the restore button. If the revision isn't to your liking, you can go back to the Revision Selector and choose a more applicable revision, or return to your latest one.

Revisions are useful for trying out features such as autosegmentation and the upcoming autotranscribers. They can also help you if you have made a bad change somewhere, your child/pet wrote their latest diary entry over your transcription, or if things go haywire for some other reason.