Diction One
Our cloud. Built for accuracy, speed, and scale. Always available, always improving. You speak. We handle everything else.
No server to run. No model to download. Open the app, subscribe, and start dictating with frontier speech models.
How It Works
- Subscribe to Diction One in the app (free trial included).
- Switch to the Diction keyboard in any text field.
- Tap the mic and speak. Audio is sent to our cloud, transcribed, and the text is returned.
- Audio is discarded immediately. Nothing is stored. Nothing is used for training.
What You Get
- Frontier speech models: the most accurate models available, hosted and maintained by us. We pick the best model for your language automatically.
- AI Enhancement: optional post-processing that cleans up your transcription. Removes filler words, fixes grammar, adds proper punctuation and capitalisation.
- 99 languages: full multilingual support. Switch languages in settings.
- Zero setup: no server, no Docker, no downloads. Works immediately after subscribing.
- Audio immediately discarded: we transcribe and it's gone. Nothing is recorded. Nothing is stored. Nothing is shared. Nothing is used to train models.
AI Enhancement
When you speak, you don't speak in perfect sentences. You say "um", pause mid-thought, skip punctuation, and occasionally trail off. That's normal.
AI Enhancement takes the raw transcription and cleans it up before inserting the text. It removes filler words like "um" and "uh", adds punctuation and capitalisation, fixes grammar, and produces text that reads like you wrote it, not like you said it.
The result: you dictate a quick thought and get a polished sentence. Dictate a paragraph and get something you'd be comfortable sending as an email.
What it does
- Removes filler words ("um", "uh", "like", "you know")
- Adds punctuation and capitalisation
- Fixes grammar and sentence structure
- Preserves your meaning and intent. It cleans up how you said it, not what you said.
What it does not do
- It does not change the meaning of your words
- It does not add content you did not say
- It does not translate between languages
- It does not store or learn from your transcriptions
How to use it
AI Enhancement is enabled by default for Diction One subscribers. You can turn it off any time in the Diction app settings if you prefer the raw transcription output.
When AI Enhancement is active, the transcription text (not the original audio) is processed by an AI model to produce the cleaned result. Audio is always discarded immediately after transcription, regardless of whether AI Enhancement is on or off.
Best For
- You want the highest accuracy without managing any infrastructure
- You dictate emails, notes, and documents and want polished output, not raw speech
- You value convenience. Subscribe and it works, everywhere, immediately.
- You have tried on-device and want a step up in accuracy