Translation & Transcription

Owned by Kyle
Last updated: May 10, 2024
4 min read
translator-banner2.png

In order to run Translation and Transcription, you must have access to AWS. The keys used to access AWS are set using the Cavrnus Administration Console under the Extensions section for the Transcribe Extension.

Create a new Extension, giving it a name and setting the extension type to Transcription. Once created, the extension can be configured. Extensions are implemented as “bots” and will join the users in the space to act as a transcriber and translater. Also, remember to Enable it in the Edit Extensions menu.

AWS Setup

Edit Extension

The Edit Extension window looks like this

image-20240314-182041.png

  • The Name is the value provided when the extension was created.

  • The Role should remain as Transcriber Extension.

  • The Screen name should be set to whatever name will be used to invite the transcriber to a Cavrnus Space.

  • The Profile Picture URL is the URL that should be used to represent the Transcriptor Bot in the Users List.

Configuration Variables:

"AWS_ACCESS_KEY_ID":"", // https://console.aws.amazon.com/iam
"AWS_SECRET_ACCESS_KEY":"", // https://console.aws.amazon.com/iam
"AWS_REGION":"", // https://docs.aws.amazon.com/general/latest/gr/transcribe.html and https://docs.aws.amazon.com/transcribe/latest/dg/what-is.html#tsc-regions

Note it is important to set the AWS_REGION as the custom vocabularies and vocabulary filters are configured per REGION.

There is a list for each language for a custom vocabulary and vocabulary filter. Note that each language must be configured individually. The keys in the Configuration menu appear similar to

"vocabularyName-en-US" with the key value being the name used to create the vocabulary on the AWS console. The transcription options can be found on AWS by searching for Amazon Transcribe.

Each language also supports a custom filter with the keys in the Configuration appearing similar to

"vocabularyFilter-en-US". More information on custom vocabulary and vocabulary filters can be found at

https://docs.aws.amazon.com/translate/latest/dg/what-is-languages.html

https://docs.aws.amazon.com/transcribe/latest/dg/custom-vocabulary.html

https://docs.aws.amazon.com/transcribe/latest/dg/vocabulary-filtering.html

https://docs.aws.amazon.com/transcribe/latest/dg/vocabulary-filter-using.html

https://docs.aws.amazon.com/transcribe/latest/dg/vocabulary-filter-create.html

In addition, Cavrnus transcription and translation supports “translateFormality”, “translateBrevity” and the “translateFilterProfanity” settings available with AWS Transcribe. See the following links for more information about these features :
https://docs.aws.amazon.com/translate/latest/dg/customizing-translations-formality.html
https://docs.aws.amazon.com/translate/latest/dg/customizing-translations-brevity.html
https://docs.aws.amazon.com/translate/latest/dg/customizing-translations-profanity.html

Two Cavrnus specific values are also available. These should usually be left empty and the default value used. These values are for the “periodicSendRate” and “idleConclusionDuration”.
periodicSendRate is used to update the Cavrnus client when a speaker is continuously speaking. It will force a visual update at a rate in seconds of the periodicSendRate value. This is useful when a speaker is speaking continuously as without this value, the transcription and translation will not appear until the user finishes speaking.
idleConclusionDuration is used in the case that AWS does not detect the user is done speaking. If the user has been silent for this period, Cavrnus informs AWS that the current transcription stream is complete and it will process the information without waiting for more.

Amazon Transcribe Menu

 

This is an image of the Amazon Transcribe menu. The individual configuring the settings will most likely only need to access “Custom Vocabulary” and “Vocabulary Filtering”. Both of these menus allow the user to set up a profile that is used to add options to. It is this profile’s name that is used with the “vocabularyName” and “vocabularyFilter” respectively.

 

 

 

 

 

 

 

 

 

Custom Vocabulary

Custom vocabulary can be created on the AWS console or via File Upload or import. For a small number of modifications, it can easily be done in the AWS console. For a longer list, download the template, edit it and upload it back to AWS.

In the image following, note the Name field is what is used to identify the vocabulary profile in the Cavrnus Transcriptor Configuration. Note that the Language field must match the language being specified in the Cavrnus Transcriptor Configuration. For example, vocabularyName-en-US as shown below. Note that the Add Row option can be used to add vocabulary values one by one vs editing a file offline and uploading it to AWS.

Other references:

https://docs.aws.amazon.com/translate/latest/dg/what-is-languages.html

https://docs.aws.amazon.com/transcribe/latest/dg/custom-vocabulary.html

Vocabulary Filtering

The vocabulary filter is more simple than the template used for custom vocabulary. In this case, a file is created with a list of words for a particular language that should be filtered out. Each word should reside on its own line. These words will be replaced by ‘***’ in the transcripted text. Remember that each language would need to set the vocabulary filter.

 

Other references

https://docs.aws.amazon.com/transcribe/latest/dg/vocabulary-filtering.html

https://docs.aws.amazon.com/transcribe/latest/dg/vocabulary-filter-using.html

https://docs.aws.amazon.com/transcribe/latest/dg/vocabulary-filter-create.html

Desktop Usage

 

 

 

 

Setting Available Languages

In the Organization Settings available through the admin console, there is a Feature called “Translation Settings.” It allows the administrator to set the available speaking and transcribed languages that are visible in the application and web client drop downs. Check all languages that the user wishes to be available in the list for users to select.