Updating models and voices for your Watson Speech services

After you install the Speech services, you might need to add or delete Speech to Text models or Text to Speech voices. You can update the installed models and voices by editing the custom resource.

Permissions you need for this task:: You must be an administrator of the Red Hat® OpenShift® project to manage the cluster.

Updating the installed models and voices with the custom resource

You can update the Speech to Text models and Text to Speech voices that are installed in your deployment at any time. You add or remove models and voices by editing your custom resource. Only models and voices for which the enabled property is true are installed. Be careful to avoid unintentionally removing models and voices by changing their enabled properties to false.

For example, the default custom resource installs US English models and voices. The following abbreviated example shows the installed models and voices that appear if you edit this custom resource. Note that the example shows only the models and voices, not the remaining properties of the custom resource. Note also that comments are not preserved from the original custom resource that was created.

. . .
  global:
    defaultSTTModel: en-US_BroadbandModel
    sttModels:
      enUsBroadbandModel:
        enabled: true
      enUsNarrowbandModel:
        enabled: true
      enUsShortFormNarrowbandModel:
        enabled: true
      enUsMultimedia:
        enabled: true
      enUsTelephony:
        enabled: true
    defaultTTSVoice: en-US_MichaelV3Voice
    ttsVoices:
      enUSAllisonV3Voice:
        enabled: true
      enUSLisaV3Voice:
        enabled: true
      enUSMichaelV3Voice:
        enabled: true
. . .

To remove the US English models and voices and replace them with the available French and French Canadian models and voices, perform the following steps:

Log in to the Red Hat OpenShift cluster:
```
oc login OpenShift_URL:port
```
Use the following command to edit the custom resource:
```
oc edit watsonspeech ${CUSTOM_RESOURCE_SPEECH}
```

Modify the custom resource to uninstall the existing US English models and voices and to install the French and French Canadian models and voices instead. You enable and disable installation of a model or voice by changing the value of its enabled property to true or false. Use the defaultSTTModel and defaultTTSVoice properties to set the default model and voice to those you plan to use most often.


. . .
  global:
    defaultSTTModel: fr-FR_Telephony
    sttModels:
      enUsBroadbandModel:
        enabled: false
      enUsNarrowbandModel:
        enabled: false
      enUsShortFormNarrowbandModel:
        enabled: false
      enUsMultimedia:
        enabled: false
      enUsTelephony:
        enabled: false
      frCaBroadbandModel:
        enabled: true
      frCaNarrowbandModel:
        enabled: true
      frFrBroadbandModel:
        enabled: true
      frFrNarrowbandModel:
        enabled: true
      frCaMultimedia:
        enabled: true
      frCaTelephony:
        enabled: true
      frFrMultimedia:
        enabled: true
      frFrTelephony:
        enabled: true
    defaultTTSVoice: fr-FR_ReneeV3Voice
    ttsVoices:
      enUSAllisonV3Voice:
        enabled: false
      enUSLisaV3Voice:
        enabled: false
      enUSMichaelV3Voice:
        enabled: false
      frCALouiseV3Voice:
        enabled: true
      frFRNicolasV3Voice:
        enabled: true
      frFRReneeV3Voice:
        enabled: true
. . .

Save your changes and exit the editor.

It can take from 20-60 minutes for your installation to reflect the changes. This estimate includes time for the operator to identify the changes and for the updated models and voices to be uninstalled and installed.

Updating models and voices for your Watson Speech services

Related topics

Updating the installed models and voices with the custom resource