Audio Chat Box

Audio Chat Box: Open source speech program that reads text aloud and turns microphone input into text locally.

Audio chat box is a standalone program that communicates over standard input and output using JSON messages. It reads text aloud with Kokoro and turns speech into text with Parakeet TDT, using Silero VAD for microphone streaming. It runs completely offline and locally through sherpa-onnx, requiring no network connection, account, or API key. Programs can supply custom hotwords to prioritize specific terms such as product names during transcription. It originates from an AI chess coach and now integrates with any program that implements its JSON protocol.

Audio Chat Box at a glance

Best for: Software engineer
Runs on: Web, self-hosted, open source, Docker
Pricing: Free
Released: October 2026
Category: AI audio tool
Website: github.com/cjus/audio-chat-box

Key features

Reads text aloud with Kokoro in 54 voices
Turns speech into text with Parakeet TDT
Streams microphone input with Silero VAD
Communicates over standard input and output via JSON messages
Supports custom hotwords to prioritize specific vocabulary
Runs entirely offline and locally through sherpa-onnx

Use it for

Add local speech-to-text and text-to-speech to an AI project
Stream microphone audio for hands-free interaction
Read text aloud using one of 54 voices
Transcribe audio files or raw samples locally

Alternatives to Audio Chat Box

AssemblyAI — Industry-leading speech models deliver high accuracy and low latency for voice applications.
Kokoro-FastAPI — Generate hours of high quality speech with an OpenAI-compatible endpoint.
kokoro-onnx — Runs local TTS inference with efficient ONNX runtime performance.
Speech-to-Speech — Fully modular, swappable components with local or hosted execution options.
Pipecat — Builds real-time voice and multimodal agents with frame-level control.
Macvoice — Runs locally on Apple Silicon in under 2 GB with fast response times.
FluidAudio — Runs local audio AI models on Apple Neural Engine.
Babytalk — Runs entirely on-device with no cloud or internet required.
browser-whisper — Transcribes audio and video locally with zero servers involved.
WeNet — Achieves state-of-the-art results on public speech datasets.
Vapi — Provides scalable orchestration for voice AI agents with sub-500ms latency.
Willow — Fast, private, open source voice control using local hardware.

All Audio Chat Box alternatives