Languages and processing models
Choose the languages you speak and understand the difference between remote and local transcription.
Recording languages
On Mac, open Settings → Dictation → Recording languages. On iPhone or iPad, open Settings → Preferences → Languages.
- Choose Auto Detect as the recommended starting point, or select up to three languages you regularly speak.
- If detection chooses the wrong language, select only the languages you regularly speak instead of using Auto Detect.
- If you switch languages in one sentence, include both languages in the selection.
If a result contains unrelated words after you stop speaking, check the microphone for background noise and choose your language explicitly before trying again.
Remote processing
Remote transcription is the recommended default and generally provides the best accuracy. Audio is sent to Monologue’s processing service for transcription and smart formatting.
Local processing on Apple silicon Mac
Local mode transcribes the raw audio on your Mac and can produce a raw transcript without a network connection. Its accuracy may be lower than remote transcription.
Smart cleanup and context-aware formatting still use Monologue’s online service. Without a network connection, the local model can produce raw transcription but not the same polished result you receive from smart formatting.
Download the local model
Select Local under Settings → Dictation → Transcription source, then download the model when prompted. Keep the app open until the status shows that the model is installed. You can delete the model from the same control when you no longer need it.
The model’s storage location is managed by Monologue’s transcription library and may change between releases. Download it from the app rather than copying files into a hard-coded folder from an older guide.
If a language is repeatedly wrong, report the affected transcript with the in-app 👎 action so support can inspect the exact example.