Skip to content

TTS provider (text-to-speech engine selector)

1. Overview

tts-provider is an Atlas service family in the media category. Its implementation and service-owned documentation live under services/tts-provider/.

2. Role In Atlas

Atlas uses this service according to its manifest, topology row, SOURCE settings, dependencies, and runtime data-flow declarations.

3. Tracks And Category

  • Category: media
  • Kind: virtual
  • Tracks: all, gen-ai-creative, gen-ai-eng

4. Access

  • Kong aliases: tts.localhost
  • Port variables: TTS_PROVIDER_PORT

5. Configuration

  • SOURCE variables: TTS_PROVIDER_SOURCE
  • Default SOURCE values: speaches-container-cpu
  • Available SOURCE values: speaches-container-cpu, speaches-container-gpu, chatterbox-container-gpu, chatterbox-localhost, disabled

6. Dependencies And Topology

  • Required dependencies: litellm
  • Optional dependencies: -
  • Runtime calls: -

7. Source Values

SOURCE Variable Default Values
TTS_PROVIDER_SOURCE speaches-container-cpu speaches-container-cpu, speaches-container-gpu, chatterbox-container-gpu, chatterbox-localhost, disabled

8. Runtime Integration

The manifest data-flow list declares runtime calls to -. The topology row supplies aliases and port surfaces used by the generated gateway and service references.

9. Architecture

10. Operations

Use ./start.sh to configure this service through the wizard or pass the matching SOURCE flag when the service is source-configurable. Use ./stop.sh to stop the active Atlas project.

11. Source Documentation