Ultra low-cost, low-latency voice AI agents that scale with Inworld TTS 1.5
Production-grade text-to-speech at $0.005/min. Realtime latency, support for 15 languages, enhanced timestamps, and more.
We're excited to announce that Inworld TTS 1.5 is now available on Layercode.
For teams building voice agents at scale, TTS 1.5 hits the sweet spot: fast enough for natural conversation, and affordable enough to scale in production.
Here's what we love about Inworld TTS 1.5:
-
$0.005 per minute: Inworld TTS 1.5-mini costs $5/M characters, TTS 1.5-max costs $10/M. That's 4-24x cheaper than other providers. Example: pairing Deepgram Flux ($0.0077/min) with Inworld TTS 1.5-mini ($0.005/min) on Layercode ($0.04/min) gives you a production-ready voice agent for $0.053 per minute.
-
Up to 4x faster: TTS 1.5-max targets sub-250ms latency: fast enough for true interruptibility and natural back-and-forth conversation. At this threshold, conversations can feel incredibly fluid and responsive.
| Model | P50 Latency | P90 Latency |
|---|---|---|
| TTS 1.5-max | 200ms | 250ms |
| TTS 1.5-mini | 100ms | 130ms |
-
15 languages: English, Spanish, French, German, Italian, Portuguese, Chinese, Japanese, Korean, Dutch, Polish, Russian, plus new support for Arabic, Hebrew, and Hindi. That's a meaningful expansion for teams deploying voice agents globally.
-
Improved voice quality: 30% greater expressiveness and 40% reduction in word error rate, significantly reducing hallucinations, cutoffs, and artifacts vs. prior generations.
-
Impressive voice cloning: This release improves on TTS 1's voice cloning to make voices feel more stable and realistic.
-
Enhanced timestamps (experimental): Phoneme and viseme support for lip-sync and animation use cases. Currently experimental and English-only.
-
Two variants: TTS 1.5-max for most production workloads; TTS 1.5-mini for ultra low-latency deployments at scale.
Inworld voices have consistently ranked among the most natural in blind comparisons on Artificial Analysis and Hugging Face. In our testing, TTS 1.5 delivers the voice quality you'd expect from more expensive voice models, at a fraction of the cost.
More choice for your real-time voice AI agents
Layercode gives you flexibility to choose the right TTS model for your use case: latency, language coverage, voice quality, or cost.
Inworld TTS 1.5 is a strong option for teams building agents for scale who don't want to trade naturalness for budget. Pair it with one of Deepgram's leading STT models on Layercode's edge network for a production-ready voice pipeline with global reach.
Build with Inworld TTS 1.5 today
Inworld TTS 1.5 is available now in your Layercode dashboard. Select it from the TTS model picker under your agent settings.
For full details, check out Inworld's TTS documentation.
New to Layercode? Sign up for a developer account and get $100 in credits to build your first voice agent.
