Introduction



Voicemail transcription is a feature in PBXware which allows customers to attach the transcript of voicemails when sending out emails to users while also being able to access this using OSC. Currently, there are 2 transcription services supported for this feature:

  • Google Speech

  • IBM Watson


Configuration


This feature will be disabled by default and needs to be configured under Settings -> Voicemail -> Advanced Optionin the Transcription section. 

Any fields active in this section are mandatory and can’t be empty. Initially, only the Enable button will not be disabled:



Also, this feature needs to be enabled on Master tenant and Slave tenant. Under the Transcription section in Tenant Options, you need to set Enable Transcription to Yes.




A wide variety of languages are supported:

  • English (US)

  • English (UK)

  • Spanish

  • French

  • Bengali

  • Greek

etc.


 Depending on the selected service additional fields appear that have to be filed with the service-specific credentials.

NOTE: In order for Speech-to-Text transcription to work, accounts for IBM Watson and Google Speech are to be set up first. IBM Watson (https://www.ibm.com/cloud/watson-speech-to-text) is free of charge and Google Speech (https://cloud.google.com/speech-to-text/) service comes with a small monthly charge.



IBM Watson




If you select IBM Watson under Service following additional options are available:

  • URL - Endpoint with which system will communicate

  • API Key - API Key used for authentication

  • Language - Which language the transcription service expects (in which language audio file should be) (Select box)


Google Speech





If you select Google Speech under Service following additional options are available:

  • API Key - API Key used for authentication

  • Language - Which language the transcription service expects (in which language audio file should be) (Select box)



Service credentials acquisition


IBM Watson


In order to get the credentials for IBM Watson’s speech-to-text service, an IBM Cloud account is required (a free account works as well but limits the number of requests that can be done). After logging into your IBM Cloud account select Watson from the menu that opens after clicking on the button in the top left corner:



On the page that opens under Watson Services search for Speech to text. From the offered services select Speech to text then press the Start free plan.




On the next page under the Service Name enter the project name. Afterwards select the resource group, if multiple are available, as well as the deployment region. You can also add an optional tag to your project.



After these options are filled in scroll to the bottom of the page, there you will select your pricing plan and press Create to create your project.

Once the project is created go to the credentials page of the project.



There will be a set of auto-generated credentials to be used or new credentials can be made by pressing the New Credentials button.



In case new credentials are created the following window will appear where you enter the credentials name and role.
Optionally the service credential ID and configuration parameters can be added.
 Once the mandatory fields are set press
 Add.


Once a service credential is open for view, only 2 fields are needed for voicemail transcription:apikey and url.




Google Speech


Log into google cloud console with an existing account. From the menu on the left select APIs and Services:



Either click Library or ENABLE APIS AND SERVICES:







From here either manually find the Google Cloud Speech API or using the search field and select it. On the Google Cloud Speech API page click the enable button:




After enabling the api press the MANAGE button that replaced the enable button. On the manage page go to the Credentials section and press Create credentials and select API key:




A popup window will appear containing the generated key press the copy icon to copy the API key needed for this feature. Some restrictions can be placed on individual keys (IP address, ios/android apps..) need be placed on the key by pressing the RESTRICT KEY button.




Already created keys can be accessed from the same page in the API keys section (the site might need to be refresh in order for keys to display). IP restriction is recommended and can be set and can be set by selecting IP addresses in the Key restriction section of the API key options. Addresses are added separately by typing them in the IP address input field and pressing enter confirm each one. Already added addresses will display above the input field where they can be edited or removed by deleting the content of the row or pressing the appropriate button.




Usage


The transcript can be added in the Voicemail E-mail Template as shown below.


In order to make this feature easier to be used the default HTML template got updated, adding an additional row with the transcript. That row will be omitted if no transcript is available. An additional option has been added to all e-mail templates that enables the template to be reset to the default one. This reset will affect all the templates for the resource it is used on (e.g.: if voicemail e-mail templates get to reset it will affect the plain text template, HTML template, and pager template). A warning message will pop up where confirmation for the reset is required in order to evade accidental resets.



To see the transcript in OSC click the speech bubble left of the message number:


NOTE: Transcription setup is available system-wide and there it is not available per Tenant at the moment.