Instead, Echos is a headless kernel: a powerful, API-driven engine designed from the ground up to serve as the core for autonomous music production systems. It provides the foundational building blocks for anyone to create their own AI-powered music tools, services, or even full-fledged DAWs.

Our vision is to empower AI to act as a complete music producer. Through a comprehensive API, an AI agent can command the entire workflow:

• Project Management: Create, save, load, and configure projects.
• Composition & Arrangement: Add tracks, instruments, and plugins; write MIDI notes and arrange clips.
• Sound Design: Tweak synth parameters, configure effects chains, and create automation.
• Mixing & Mastering: Control faders, panning, sends, and buses to achieve a perfect mix.

From a single prompt, an AI can orchestrate a complete song from scratch.

Echos is designed to make this vision a reality.

• Our goal is to expose every function of a traditional DAW through a clean, programmatic toolkit. What you can click in a GUI, an AI can call via our API.
• The toolkit is constantly expanding, giving AI agents ever-increasing creative and technical capabilities.

• The core logic is decoupled from the audio backend via an event bus. This allows for ultimate flexibility and ensures the architecture is future-proof.
• We currently ship with a high-performance backend using Spotify's Pedalboard, but new engines can be easily integrated.

This example shows an AI performing a full, albeit simple, production workflow.

git clone https://github.com/linzwcs/echos.git
cd echos
pip install -e .
pip install pedalboard sounddevice

import sys
from pathlib import Path
import json
from echos.core import DAWManager
from echos.backends.pedalboard import (
    PedalboardEngineFactory,
    PedalboardNodeFactory,
    PedalboardPluginRegistry,
)
from echos.core.plugin import PluginCache
from echos.core.persistence import ProjectSerializer
from echos.facade import DAWFacade
from echos.services import *
from echos.agent.tools import AgentToolkit, tool



def initialize_daw_system():

    print("\n" + "=" * 70)
    print("Initializing DAW system...")
    print("=" * 70)

    plugin_cache = PluginCache()
    plugin_registry = PedalboardPluginRegistry(plugin_cache)
    engine_factory = PedalboardEngineFactory()
    node_factory = PedalboardNodeFactory()
    serializer = ProjectSerializer(node_factory, plugin_registry)

    manager = DAWManager(
        serializer,
        plugin_registry,
        engine_factory,
        node_factory,
    )

    services = {
        "project": ProjectService(manager),
        "node": NodeService(manager),
        "transport": TransportService(manager),
        "editing": EditingService(manager),
        "history": HistoryService(manager),
        "query": QueryService(manager),
        "system": SystemService(manager),
        "routing": RoutingService(manager),
    }

    # Create the Facade
    facade = DAWFacade(manager, services)

    print("✓ DAW system initialized successfully")

    return facade, manager

facade, manager = initialize_daw_system()
toolkit = create_agent_toolkit(facade)

result = toolkit.execute("project.create_project", name="Pop Song")
print(f"  ✓ {result.message}")

For a detailed example of how to create a song, please refer to demo/song_create_demo.py.

Notice: Before creating a song, you need to install the required VST3 instrument plugins.

Our journey is about progressively expanding the AI's capabilities across the entire production lifecycle.

• 🔄 Solidify the API for core project, track, and plugin management.
• 🔄 Improve VST3/AU hosting reliability.
• 🔄 Expand the Toolkit: Add more essential tools for editing, mixing, and routing.

• 🎯 Implement robust audio and MIDI file import/export.
• 🎯 Develop a comprehensive automation system via the API.
• 🎯 Introduce more complex routing options like sends and groups.

• 🔮 Build higher-level tools for common tasks (e.g., "create a drum beat," "master for Spotify").
• 🔮 Introduce a simple UI for visualizing the AI's actions.
• 🔮 Explore ML-powered tools for intelligent decision-making.

Echos presents a unique opportunity to define the future of music production. We are looking for passionate contributors to help us build out the AI's arsenal of creative tools. Every function you expose is a new power you grant to the autonomous producer.

Whether you're defining the contract for a foundational component, crafting a clean, robust programmatic API, architecting a complex and scalable new feature, or pioneering research into autonomous agents and novel AI architectures for composition, your contribution is invaluable. Join us in building the core infrastructure that will power the next generation of musical creativity.

To join us, please reach out to zhilin.nlp@gmail.com.

This project is licensed under the Apache 2.0 License - see the LICENSE file for details.