Provider Selection Guide
Choose the right STT, LLM, and TTS for your use case
A decision framework for selecting STT, LLM, and TTS providers in LiveKit's modular pipeline. Compare latency, accuracy, cost, and language support across providers with concrete recommendations for common use cases.
What You Build
Select and configure the optimal provider stack for a given business scenario.
Prerequisites
- →Course 1.1
Pipeline selection & latency budgets
20mHow LiveKit's modular pipeline lets you choose each provider independently. Understand the latency budget (STT + LLM + TTS target: ~500ms) and how streaming overlap reduces perceived latency.
STT, LLM & TTS provider comparison
25mSide-by-side comparison of providers at each pipeline stage: Deepgram vs Whisper vs Azure for STT, GPT-4o vs Claude vs Gemini for LLM, Cartesia vs ElevenLabs vs PlayHT for TTS. Latency, accuracy, cost, and language support.
Cost analysis & use case recommendations
15mCost breakdown for typical 5-minute conversations across budget to premium stacks. Concrete recommendations for customer service, healthcare, education, and entertainment use cases.
What You Walk Away With
Ability to evaluate and select STT, LLM, and TTS providers based on latency, accuracy, cost, and use case requirements.