Small, fast, free: an open recipe for a sub-100M TTS. Code, data, weights. The PoC trained to a working model in a few hours on one RTX 3090, and inspired a wave of open-weights TTS releases.

onnxopen-sourcespeech-synthesistext-to-speechtts

toy-local-multi-agent-swarm

Repo · 12 stars · Python

▲ 0

Experiment: Watch a swarm of LLM agents run on your own machine, explore different swarm topologies, and nudge or redirect any agent in real time.

supertonic-voicelab

Repo · 10 stars · Python

▲ 0

40 new voices for Supertonic-3 TTS + labeled, bounded sliders to make your own. No retraining, no GPU.

llm-sandbox

Repo · 5 stars · JavaScript

▲ 0

Talk to a tiny sassy robot and she builds your voxel world. Fully local: Gemma 4 + Whisper + Supertonic.

open-swarm-security

Repo · 0 stars · JavaScript

▲ 0

Interactive article on keeping swarms of parallel LLM agents in scope: UX, architecture, threat model, controls

About

Francisco Carlos Erra works as an independent Agentic AI Engineer from Buenos Aires, Argentina. Francisco is known for ttslibre, an open recipe for a small, fast text-to-speech model under 100M parameters, with code, data, and weights released publicly. Francisco's current work centers on local AI agents and voice tools, including a multi-agent swarm that runs on a personal machine and a fully local voice project built on Gemma 4, Whisper, and Supertonic.