Say It Like
Clone any voice. Generate speech. Entirely on-device.
Start new thread
Ashwary

Say It Like - Clone any voice. Generate speech. Entirely on-device.

by•
Say It Like is a native macOS voice cloning and text-to-speech app built entirely in Swift. - Clone a voice from any audio file, mic, or system audio - Generate speech in that voice from any text, with live streaming output - Works fully offline. Generation runs on Mac, nothing leaves the device For the more technical folks, no Python runtime bundled, no server calls, no background processes phoning home. Runs on MLX. I'm a contributor to the MLX audio libraries that power this app.

Add a comment

Replies

Best
Ashwary
Maker
📌
I built Say It Like because I wanted voice cloning I could actually trust. Yes, there are many other TTS apps out there. Most bundle a Python runtime and a pile of dependencies, or wrap a web UI in an Electron shell and call it native. Not many of them use MLX, or stream audio as it generates, and none treat the OS as something worth integrating with properly. So the constraints I set: native Swift only, MLX for inference, streaming output, no Python runtime, real OS integration. That meant contributing to the MLX audio libraries myself to get the behaviour I needed. The foundation this is built on is not a black box I am working around. The result is an app that feels like it belongs on your Mac. Fast to start, fast to generate, and wired into the system in ways the alternatives are not. If you are into the technical side, happy to go deep in the comments.