Thanks — you’re asking for alternatives to Cartesia and Inworld with similar pricing and good latency. Here are practical competitors to consider (price tier and latency notes included): • Deepgram (Aura-2) — Competitive pricing (~$0.018/min Build, ~$0.0162/min Scale) and low latency (~100 ms). Good balance of cost and real-time performance. • Rime (Mist / Arcana) — Cost-competitive (around $0.012–$0.018/min on Scale for Mist; Arcana a bit higher). Real-time/streaming support through LiveKit Inference; exact latencies not published. • xAI Text-to-Speech — Very low cost ($0.0025/min) but latency details aren’t published; worth testing for your region/workload. • ElevenLabs (Flash v2.5) — Higher cost (examples: $0.090/min Build, $0.036/min Scale) with good quality and ~150 ms reported latency; useful when voice quality matters over absolute cost. • Cartesia Sonic 3 (reference point) — Low latency (~90 ms) and premium voice quality; you may compare other vendors against this. Notes and next steps: • Latency varies by model, region, and network; I recommend testing 1–2 models in your target region to measure real-world latency and voice quality. • If you tell me which regions, languages, or voice-quality vs cost tradeoffs matter most, I can narrow this list and suggest which to test first. Sources: Rime TTS | LiveKit Documentation | Pricing | LiveKit | Build Your First AI Voice Agent in Python: Complete Tutorial | LiveKit | Build Your First AI Voice Agent in Python: Complete Tutorial | LiveKit, with interactive elements