Run with Helm Charts

A Helm Chart packages Kubernetes resources for easy configuration and deployment. The full Speech to Text application can be deployed with a single command. This is especially useful for deploying the customization service that has more complex configuration requirements.

The Helm charts can be found at IBM/ibm-watson-embed-charts.

Runtime Only

Packages the Speech to Text runtime serving a set of pretrained models.

The ibm-watson-stt-embed-runtime chart installs the following:

  • STT runtime Deployment
    • HAProxy container used for TLS termination
    • The set of enabled models are downloaded and configured via initContainers
  • Service to expose the Deployment
  • TLS secrets with a self-signed CA and the server certificate

Runtime and Customization

Packages the Speech to Text service with a runtime for serving a set of pretrained models and customization for creating a new custom language model for a specified base model. Custom language models can add domain specific corpus, custom words, or custom grammar.

The ibm-watson-stt-embed chart installs the following:

  • STT customization Deployment
    • initContainer that will create the required PostgreSQL database
  • STT runtime Deployment
    • HAProxy container used for TLS termination
  • Job(s) to upload the set of enabled STT models into storage
  • Services to expose the Deployments
  • Secrets for PostgreSQL and model storage
  • TLS secrets with a self-signed CA and the server certificate

After deploying the STT helm chart, see the customization example for an example of using customization.