ASR → TTS echo pipeline. Speak and hear it back.
- Device: Parakeet ASR + Kokoro TTS (CoreML E2E)
- Simulator: Parakeet ASR + Apple built-in TTS
cd Examples/iOSEchoDemo
xcodegen generate
open iOSEchoDemo.xcodeprojSet your signing team in Xcode, build and run.
Models (~500 MB) download from HuggingFace on first launch.
- Voice activity detection (Silero VAD)
- Force-cut at 5s with system message
- Adaptive echo prevention (cooldown based on TTS audio duration)
- Diagnostics view (CPU, memory, VAD level)