izwi-ai/izwi

Voice AI runtime. Local first transcription, speaker diarization, TTS, and voice cloning with an OpenAI compatible API.

What it solves

Izwi provides a local-first environment for voice AI, eliminating the need for cloud services or API keys. It allows users to run speech-to-text, text-to-speech, and chat models entirely on their own hardware to ensure data privacy.

How it works

It operates as a desktop application, web UI, CLI, and local inference server. It manages the downloading and execution of various model families for ASR (Automatic Speech Recognition), TTS (Text-to-Speech), and LLM chat. The system also exposes OpenAI-compatible API routes, allowing other applications to use its local inference capabilities.

Who it’s for

It is designed for users who want to perform audio workflows—such as transcription, voice cloning, and real-time voice conversations—locally on their machines for privacy or offline access.

Highlights

  • Comprehensive Audio Suite: Supports real-time voice conversations, long-form studio projects, speaker diarization, and forced alignment.
  • Local-First Privacy: Inference data stays on the local machine, with no required cloud connectivity.
  • OpenAI Compatibility: Provides /v1 API routes for chat completions and audio tasks.
  • Broad Model Support: Integrates multiple model families including Qwen3, Whisper, Kokoro, and Gemma.

Related

  • Project
    jamiepine/voicebox

    An open-source, local-first AI voice studio for voice cloning, speech generation, and system-wide dictation, featuring an MCP server for AI agent integration.

  • Project
    Quantatirsk/qwen3-asr

    Qwen3‑ASR is a self‑hosted speech‑to‑text server that wraps Qwen3‑ASR models (0.6 B & 1.7 B). It auto‑selects a GPU‑vLLM backend or a CPU‑Rust backend, offers OpenAI‑ and Alibaba‑compatible HTTP/WebSocket APIs, supports speaker diarization, VAD‑based segmentation, batch inference, and can be deployed via Docker, custom GPU images, or offline tarballs. MIT‑licensed.

  • Project
    isair/jarvis

    A 100% private, offline AI voice assistant that provides conversational awareness and unlimited local memory, integrating with thousands of tools via MCP.

  • Project
    debpalash/VoiceStudio

    A local-first audio production suite for voice cloning, video dubbing, and long-form audio generation using a variety of open-source TTS and ASR engines.

  • Project
    akshayaggarwal99/jarvis-ai-assistant

    An open-source voice dictation tool that converts speech to text anywhere on your system, supporting local transcription and LLM post-processing for privacy and speed.