Demos

Full-featured speech app with CLI and SwiftUI. Transcription, TTS, VAD, diarization, speech enhancement — all models accessible from one interface.

Minimal echo pipeline for iOS — tap to talk, speech is transcribed and echoed back via Kokoro TTS. Real-time VAD visualization.

Full-duplex speech-to-speech using PersonaPlex 7B. Speak naturally, get spoken responses — like a conversation with an AI. 18 voice presets.

Install & Run

brew install soniqo/tap/speech

Usage examples:

speech transcribe file.wav
speech speak "Hello world"
audio server

Open the project in Xcode and run on a physical device (iOS 18+). Simulator is not supported for on-device inference.