WriteVoice is voice dictation that formats what you said for where you are sending it. A custom keyboard makes it available inside any app, and the tone adapts to the destination – the same spoken thought becomes a LinkedIn post, a WhatsApp message or a Gmail draft with different register in each. Around it sit rewriting controls (shorten, expand, fix grammar, emojify), meeting transcription and summarisation, translation into more than 30 languages, a Whisper Mode for dictating quietly, and sync across iOS, Mac and web.
App-aware tone is the genuinely good idea. Dictation has been technically solved for years and is still barely used, because the gap is not accuracy – it is that spoken language is not written language, and raw transcription of what you said reads badly everywhere. Knowing that a message to a colleague on WhatsApp wants different shaping from a post on LinkedIn is the missing step, and Whisper Mode addresses the other real barrier, which is that most people will not dictate where anyone can hear them.
The 98%-plus accuracy claim across accents and technical jargon is exactly the figure to be sceptical of – accuracy on accented speech and domain vocabulary is where transcription reliably degrades, and a single headline number covering both is doing a lot of work. Test it on your own voice and your own jargon before relying on it. The product also runs iOS, Mac and web only, so Android and Windows users are out entirely, and no rates are published beyond a single subscription covering every device. It is built by an individual founder.
