Palabra, a real-time speech AI company, has ranked #1 for latency on Coval's independent, open-source text-to-speech benchmark, posting 104 milliseconds — roughly twice as fast as the nearest competitor — with a 6% word error rate. The benchmark measures Time to First Audio (TTFA) using pinned, versioned datasets, and word error rate is calculated by transcribing each provider's synthesized audio with a fixed ASR model. Palabra outpaced ElevenLabs, Cartesia, and other major players. Many voice AI systems still operate at around 500 milliseconds of latency or more. Brooke Hopkins, Founder and CEO of Coval, said humans respond in around 400 milliseconds, and systems slower than that feel unnatural. Palabra says the result reflects a new TTS architecture that achieves 35ms of time-to-first-audio before network overhead, paired with production infrastructure built to hold that speed at scale. Artem Kukharenko, CEO & Co-founder of Palabra AI, noted that latency above 200 milliseconds introduces a noticeable delay that disrupts the natural flow of communication.