MuseKnowledge™ Application Speech: Requirements and Configurations
The MuseKnowledge Application embeds speech technology for the following features:
- Speech to text, which allows users to insert text into the inputs of the application, like the main search form, just by speaking to a microphone. The spoken text is recognized and it is automatically written in the selected input.
- Text to speech, which allows the results metadata to be read into natural-sounding speech.
The speech functions in Muse are currently implemented using the Microsoft Cognitive Services, for which subscription is required. To get started, a new resource must be created into the Microsoft Azure Portal, of type Cognitive Services - Speech. All API requests to Azure services require an endpoint URL and a read-only key for authenticating access.
Documentation about the Azure Speech Service with information for creating the service is available here:
https://learn.microsoft.com/en-us/azure/cognitive-services/speech-service/
When done creating the Speech resource, copy the API Keys and region details to be configured in Muse. The API Keys and URLs are available in the Keys and Endpoint section in the Azure Portal:
Speech to Text
Muse Speech to Text is available starting with MuseKnowledge Application version 7.4, release date October 2019.
Configurations
The Speech to Text functionality should be already enabled in the MuseKnowledge Application, a microphone icon should be visible in all application search inputs. If the microphone symbol is not displayed in the search inputs, the following configurations must be checked:
- Access the Muse Console for Applications Administration (MCAA), select the Muse Application and access the Application General Settings left menu item, Interface Options, Functionality tab and make sure Enable Speech to Text is enabled and the selection integration method is Microsoft Azure.
- Edit the ${MUSE_HOME}/web/www/logon/${APPLICATION_ID}/lib/microsoft/microsoft.cognitiveservices.speech.sdk.jsp file on disk, and fill in the API key and region values in the subscriptionKey and region variables.
Text to Speech
Muse Text to Speech is available starting with MuseKnowledge Application version 8.6, release date February 2023.
More details about Muse Text to Speech are available in the MuseKnowledge™ Application 8.6 Release article.
Configurations
The Text to Speech functionality should be already enabled in the MuseKnowledge Application, a Play button should be visible in the results page, for each result, in footer section of the result. If not, check in the Search Options panel that the Show Speech Controls flag is set to Yes. If there is no Show Speech Controls flag, the following items must be verified, enabled and configured:
- Access the Muse Console for Applications Administration (MCAA), select the Muse Application and access the Application General Settings left menu item, Interface Options, Functionality tab and make sure Enable Text to Speech is enabled and the selection integration method is Microsoft Azure.
- Edit the ${MUSE_HOME}/web/www/logon/${APPLICATION_ID}/lib/microsoft/microsoft.cognitiveservices.speech.sdk.jsp file on disk, and fill in the API key and region values in the subscriptionKey and region variables.
- Check that the DetectLanguage Muse Source Package (SP) is installed in the application and configured with the MS Azure Translator API key. Edit its configuration in MCAA, (select the Source Package and access Edit Source Advanced Configuration) and in the Custom Parameters place the API key as obtained from the Azure portal as value for the API_KEY metavariable.
Access the below links to view recordings of this functionality, one taken while in the English interface of the MuseKnowledge Application and for an English result, and the second taken while in the Arabic translated interface, for an Arabic result.