Observed arrival · 2026-10-25
TTS for Voice Agents: a latency-first benchmark
A public comparison of text-to-speech APIs for real-time voice agents, organized by latency and cross-referenced with quality, accuracy, and price.
○Open
⊠Login
$Paid
†Ads
✦Pretty
●Pro
◎Niche
◉Human
⚑Risk
ƒJS
- For
- Teams choosing text-to-speech APIs for voice agents
- Worth noticing
- Its method keeps median and P95 latency separate and treats benchmark rankings as non-combinable.
Field notes
The page reports 119 models from 48 providers, with 32 assigned a speed tier using Coval measurements. Its methodology distinguishes typical-turn median latency from slower-turn P95 latency, treats gaps under 10 ms as ties, and keeps results from separate benchmarks uncombined. A language section reports where Artificial Analysis or Speko actually tested models, rather than treating missing entries as negative results.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
○OpenPublic substance visible
✦PrettyNotable craft visible
◎NicheUnusually specific use
One card from the complete issue