{"slug": "cekura-tests-26-speech-models-and-finds-speed-does-not-buy-naturalness", "title": "Cekura tests 26 speech models and finds speed does not buy naturalness", "summary": "Cekura published a public benchmark of 26 text-to-speech models on October 19th comparing latency, reading accuracy, naturalness and interruption behavior for voice agents, according to results shared by co-founder and CEO Tarush Agarwal. Inworld TTS-2 Flash posted the fastest full-sentence latency at a median 46 milliseconds before speaking, while Fish Audio Drama 3 led the automated naturalness estimate despite ranking well behind on speed. Cekura's dual role as evaluator and vendor gives buyers reason to examine the methodology alongside the leaderboard.", "body_md": "# Cekura tests 26 speech models and finds speed does not buy naturalness\n\n**Tarush Agarwal's benchmark puts Inworld TTS-2 Flash first for full-sentence latency, while Fish Audio Drama 3 leads an automated naturalness estimate.**\n\n        By [Ryan Merket](https://runtimewire.com/author/ryan-merket)\n        · Published \n\nPrimary source: [X](https://x.com/tarush_agarwal_/status/2112208348720570405)\n\n## Why it matters\n\nVoice-agent builders face tradeoffs among response time, pronunciation and interruption handling. Cekura's benchmark makes those choices easier to inspect, while its role as both evaluator and vendor gives buyers reason to examine the methodology alongside the leaderboard.\n\n[Cekura](https://www.cekura.ai/) published a public benchmark of 26 [text-to-speech](https://runtimewire.com/models/huggingface/nithu-text-to-speech-b7d77dbaa33f9d81) models on October 19th, comparing latency, reading accuracy, naturalness and interruption behavior for voice agents. The test's fastest model began speaking after a median 46 milliseconds; the model with the highest estimated naturalness ranked well behind it on speed, according to [the results](https://benchmarks.cekura.ai/text-to-speech) shared by co-founder and CEO [Tarush Agarwal (@tarush_agarwal_)](https://x.com/tarush_agarwal_).…\n\nWhen you purchase through links in our articles, [we may earn a small commission](https://runtimewire.com/affiliate-monetization-standards). This doesn’t affect our editorial independence.", "url": "https://wpnews.pro/news/cekura-tests-26-speech-models-and-finds-speed-does-not-buy-naturalness", "canonical_source": "https://runtimewire.com/article/cekura-text-to-speech-benchmark-26-models", "published_at": "2026-10-19 17:24:54+00:00", "updated_at": "2026-10-19 20:44:58.800560+00:00", "lang": "en", "topics": ["ai-products", "ai-tools", "artificial-intelligence", "natural-language-processing"], "entities": ["Cekura", "Tarush Agarwal", "Inworld TTS-2 Flash", "Fish Audio Drama 3"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/cekura-tests-26-speech-models-and-finds-speed-does-not-buy-naturalness", "markdown": "https://wpnews.pro/news/cekura-tests-26-speech-models-and-finds-speed-does-not-buy-naturalness.md", "text": "https://wpnews.pro/news/cekura-tests-26-speech-models-and-finds-speed-does-not-buy-naturalness.txt", "jsonld": "https://wpnews.pro/news/cekura-tests-26-speech-models-and-finds-speed-does-not-buy-naturalness.jsonld"}}