Description

The iOS app (openclaw-ios) currently does not support Talk Mode — the app cannot use the Gateway's native TTS (Text-to-Speech) and STT (Speech-to-Text) pipeline.

Expected Behavior

When connected to a Gateway with TTS/STT configured, the iOS app should support:

• STT: Record audio on iPhone → send to Gateway → Gateway processes via configured STT provider (e.g. OpenAI Whisper API / SenseVoice) → return text to agent
• TTS: Agent generates text response → Gateway calls configured TTS provider (e.g. OpenAI TTS API / CosyVoice) → stream audio back to iPhone for playback

This mirrors the macOS app's Talk Mode behavior, which already works through the Gateway's TTS pipeline.

Current State

• The iOS app connects to the Gateway via WebSocket and text chat works fine
• The app's node capabilities show caps: [] and commands: [] — no audio/voice capabilities are registered
• The Gateway has TTS/STT properly configured and working (tested via macOS app and Feishu voice messages)

Use Case

Being able to talk to the assistant hands-free on iPhone is a key mobile use case — cooking, driving, walking, etc. Text-only on mobile is limiting.

Environment

• iOS app version: 2026.6.1 (built from source)
• iPhone: iPhone 15 Pro Max (iPhone16,2), iOS 26.6.0
• Gateway: 2026.5.28, macOS 26.5
• TTS provider: OpenAI-compatible (CosyVoice3)
• STT provider: OpenAI-compatible (SenseVoice)