Translation service or AI model?
Service translation is the direct path: speech becomes text, then a supported translation API processes it. RediFinch includes provider choices such as DeepL, Google, and Azure, plus a compatible custom endpoint.
AI translation uses a language model you choose. It can be useful when tone and context matter, but the model can also paraphrase. Review names, figures, and meaning before sending important text.
Configure the modes
In Settings → Output, select a translation mode, choose its provider and model where applicable, and enter your own API key. Custom endpoint fields appear when you choose a custom provider.
Assign a separate push-to-talk shortcut to each output mode. For example, keep your usual shortcut for dictation and add a modifier for translation so you can switch in place.
- Configure a speech engine in Settings → Engine.
- Configure Service Translation or AI Translation in Settings → Output.
- Set a distinct shortcut for that mode, then dictate into a text field and review the translated result.
Know the data path
A local speech model can handle the transcription step offline, but translation through a service or AI provider sends the transcript to that provider. Cloud speech sends audio to the speech provider as well. Choose services with policies that fit the text you plan to dictate.