A community-built benchmark now covers most known text-to-speech (TTS) tools - software that converts written text into spoken audio - available as of May 2026. The tts-bench project on GitHub includes measured results for Windows and Mac hardware, with Linux testing currently in progress on additional hardware configurations.
The project was built to fill a gap that anyone evaluating TTS options has run into. Most comparisons are either vendor-produced demos or informal listener tests that don't tell you anything about speed, resource usage, or how a model performs on your specific hardware. Results are presented in an HTML page designed for easy cross-model comparison.
The Case for Running TTS Locally
The benchmark is focused on local TTS - models you run on your own machine rather than sending text to a cloud API. That matters for several practical reasons. Text processed locally never leaves your machine, which is relevant for sensitive documents or legal transcripts. It also removes per-character API pricing and rate limits from workflows that generate high volumes of audio output.
Local TTS quality has improved enough in the past year that the question has shifted from "can consumer hardware run a decent TTS model?" to "which model is fastest and most accurate on my specific setup?" A model that performs well on an Apple Silicon Mac may behave differently on a Windows machine with a mid-range GPU - exactly the kind of hardware-specific data this benchmark captures.
The project is a solo effort without organizational backing, so its update cadence depends entirely on one person's continued involvement. That's worth keeping in mind before building a tool selection decision around it. But as a current snapshot of the local TTS landscape, it's a more honest starting point than vendor benchmarks - and it covers hardware configurations that most formal evaluations skip entirely.