Inbound RNNoise on by default for voice runners
September 1, 2026 · release
Cloud Voice and Voice+Data runners now ship with inbound RNNoise enabled by default. Incoming caller audio is cleaned before VAD and STT, which helps multi-session pods in noisy environments without changing your agent bundle.
What changed
- Default on for voice runners — no advanced override needed on new deploys
- Dashboard toggle under Project → Voice → Advanced voice pipeline → Inbound noise suppression (RNNoise)
- CLI —
voicethere projects voice-advanced set noiseSuppression.enabled false - Data-only projects — the Voice panel (and this toggle) stays hidden; data-only runners do not run a voice pipeline
Changes apply on your next Deploy to cloud.
Try it
- Docs: Advanced voice settings
- Dashboard: VoiceThere dashboard → your project → Voice (STT / TTS) → Advanced voice pipeline
voicethere projects voice-advanced list
voicethere projects voice-advanced set noiseSuppression.enabled false