Diction One

A fine-tuned speech model with advanced audio filters, 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.

How It Works

1. Subscribe to Diction One in the app (free trial included).
2. Switch to the Diction keyboard in any text field.
3. Tap the mic and speak. Audio is transcribed and the text is returned.
4. Audio is discarded immediately. Nothing is stored. Nothing is used for training.

What You Get

• Fine-tuned speech model: custom-trained for accuracy beyond standard Whisper, with advanced audio filters applied before transcription. Hosted and maintained by us. We pick the best model for your language automatically.
• AI Companion: optional post-processing that cleans up your transcription and understands voice commands. Removes filler words, fixes grammar, adds proper punctuation and capitalisation. Tell it to rewrite, translate, or edit and it follows your instructions. Tell it who you are with Profile, set a tone per app with Tone Presets, and add your own terms with My Words. Tuned to read like a person wrote it, not like a model wrote it for you.
• Long dictations: long messages, long notes, full meetings come back complete. No truncation, no cutoff.
• Your words, uncensored: profanity passes through as you said it. No automatic sanitising or replacement with asterisks.
• 99 languages: full multilingual support. Switch languages in settings.
• Zero setup: no server, no Docker, no downloads. Works immediately after subscribing.
• On-device fallback: if the request fails for any reason, Diction automatically retries using a local model on your iPhone. Your dictation is never lost to a network issue.
• Audio immediately discarded: we transcribe and it is gone. Nothing is recorded. Nothing is stored. Nothing is shared. Nothing is used to train models.

AI Companion

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 Companion 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. And when you need more than cleanup, just tell it what to do -- say "translate to Spanish" or "make this more concise" and it follows your instructions.

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.
• Tone Presets: set Professional, Casual, Friendly, or a custom style per app. Diction remembers which tone to use where.
• My Words: add names, jargon, and product terms so both transcription and cleanup get them right.

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 Companion 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 Companion 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 Companion 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