feat: add TTS deployment to Helm chart

[hiteshwadhwani]

Summary

• Add lightning-tts and api-server-tts Kubernetes templates for TTS (Text-to-Speech) deployment
• Shared resources (redis, license-proxy, secrets) are reused — no duplicates created
• Move models.asrModelUrl to lightningAsr.modelUrl with backward-compatible fallback
• Add lightningTts.modelUrl for consistency across lightning services
• Bump chart version 0.2.2 → 0.3.0

New resources

Template	Purpose
lightning-tts/deployment.yaml	GPU TTS inference engine (port 8876)
lightning-tts/service.yaml	Internal ClusterIP service
api-server-tts/deployment.yaml	TTS API gateway (port 7100)
api-server-tts/config.yaml	TOML config with lightning_v31_base_url
api-server-tts/service.yaml	External service

Deprecation notice

models.asrModelUrl is deprecated and will be removed in a future major version. Use lightningAsr.modelUrl instead. The old field continues to work as a fallback — no action required for existing deployments.

Test plan

• helm template renders all TTS resources alongside ASR
• Shared resources (redis, license-proxy) are not duplicated
• lightningAsr.modelUrl takes priority over models.asrModelUrl
• models.asrModelUrl still works as fallback for backward compat
• Existing ASR-only values files deploy without errors

🤖 Generated with Claude Code

[entelligence-ai-pr-reviews]

EntelligenceAI PR Summary

Introduces full TTS stack support to the smallest-self-host Helm chart with new Kubernetes resources and values configuration.

• Added charts/smallest-self-host/templates/api-server-tts/config.yaml: ConfigMap for TTS API server with Redis and license proxy settings
• Added charts/smallest-self-host/templates/api-server-tts/deployment.yaml: Deployment for api-server-tts with health probes and secret injection
• Added charts/smallest-self-host/templates/api-server-tts/service.yaml: Configurable Service for api-server-tts
• Added charts/smallest-self-host/templates/lightning-tts/deployment.yaml: Deployment for lightning-tts with Redis connectivity and rolling update strategy
• Added charts/smallest-self-host/templates/lightning-tts/service.yaml: ClusterIP Service for lightning-tts
• Updated charts/smallest-self-host/templates/lightning-asr/config.yaml: ASR model URL now supports per-service override via lightningAsr.modelUrl
• Updated charts/smallest-self-host/values.yaml: New lightningTts and apiServerTts value blocks, replica scaling entries, and lightningAsr.modelUrl field with deprecation of models.asrModelUrl
• Bumped chart version to 0.3.0 in Chart.yaml

---

Confidence Score: 3/5 - Review Recommended

Likely safe but review recommended — this PR introduces a full TTS stack (ConfigMap, Deployment, Service) to the smallest-self-host Helm chart, which is a meaningful infrastructure addition that warrants human eyes even in the absence of automated review comments. The new api-server-tts/deployment.yaml with secret injection, health probes, and Redis/license-proxy configuration represents non-trivial Kubernetes resource definitions that could have subtle misconfigurations (e.g., incorrect probe paths, missing resource limits, or improper secret key references) not caught by static analysis alone. No automated issues were flagged, but the zero coverage of all 8 changed files means the heuristic analysis had no visibility into the actual template contents.

Key Findings:

• Zero of 8 changed files were reviewed by automated analysis, meaning potential issues in deployment.yaml secret injection patterns, config.yaml Redis/license-proxy settings, or service.yaml port configurations are entirely unverified.
• The api-server-tts/deployment.yaml introduces secret injection for a TTS service — secret key names and mount paths must exactly match the referenced Kubernetes Secret objects, and any mismatch would cause pod startup failures silently misattributed to application errors.
• New Helm chart additions should be validated for correct values.yaml conditionals (e.g., if .Values.tts.enabled) to ensure the TTS stack can be cleanly disabled without leaving orphaned resources, a pattern that cannot be verified without seeing the template logic.
• The PR achieves a coherent and well-scoped goal of adding TTS deployment support, and the structure (separate config/deployment/service files under api-server-tts/) follows good Helm chart organization conventions.

Files requiring special attention

• charts/smallest-self-host/templates/api-server-tts/deployment.yaml
• charts/smallest-self-host/templates/api-server-tts/config.yaml
• charts/smallest-self-host/templates/api-server-tts/service.yaml
• charts/smallest-self-host/values.yaml