There isn't a comprehensive leaderboard or systematic latency comparison data for common TTS providers readily available. However, here are some insights and tools that can help: • Specific Latency Data: ◦ PlayAI TTS: 350ms ◦ Google TTS: Described as "fast" but with a "robotic-sounding" voice ◦ Tortoise TTS: Produces "amazing-sounding" audio but takes ~20 seconds to generate a single sentence • Testing Infrastructure: LiveKit supports a wide range of TTS providers for comparison, including AWS Polly, Azure TTS, Google, ElevenLabs, and more. This infrastructure allows for systematic testing and comparison. • Metrics Collection: LiveKit's framework includes tools to measure key latency indicators such as Time to First Byte (TTFB), total processing time, and audio duration. This enables detailed performance analysis. • Practical Comparison Tools: LiveKit provides examples of switching between providers (e.g., Rime, ElevenLabs, Cartesia, PlayAI) for direct comparison testing. While a pre-built leaderboard isn't available, LiveKit's infrastructure and metrics collection capabilities can be used to generate detailed latency comparisons tailored to your needs. Sources: tests/test_stt.py | README.md | README.md | Live Conversations with AI using ChatGPT and WebRTC