Self-host an OpenAI-compatible speech API for local transcription, translation, and TTS with Speaches

Use Speaches when an agent stack expects OpenAI-style audio endpoints but you want a self-hosted speech backend for transcription, translation, and text-to-speech instead of a hosted API.

Prerequisites

Docker or Python-based deployment environment, CPU or GPU runtime, supported speech models, and any client or agent stack that can call OpenAI-compatible audio endpoints.

Installation

Requirements and caveats from upstream:

• Deployable via Docker Compose / Docker

Basic usage or getting-started notes:

• See the documentation for installation instructions and usage: speaches.ai

• Realtime API

• Source: https://github.com/speaches-ai/speaches

• Extracted from upstream docs: https://raw.githubusercontent.com/speaches-ai/speaches/HEAD/README.md

Documentation

• https://speaches.ai

Source

• Agent Skill Exchange