Murmure converts speech into text using a neural speech-recognition model running on the user’s computer. It supports multilingual dictation, configurable keyboard shortcuts, custom vocabulary and text-formatting rules. Optional connections to local or remote language models can translate or revise the resulting text.