VivaDictaVivaDicta

Translation

Speak in any language and get translated text — automatic or on-demand.

VivaDicta can automatically translate your speech into another language. Speak in any language and get the translated text - in one step. There are two ways to do it, and the right one depends on which transcription provider you use.

Two Approaches

  • Native translation - the transcription provider itself does the translation in the same API call. Available with Soniox, Gladia, and Speechmatics. No AI provider required, no extra round trip.
  • AI translation - the transcription is sent to your configured AI provider with a translation preset. Works with any transcription provider and any AI model.

Looking for translation while you speak (live, not after the fact)? See Live Translation.

Native Translation (Soniox / Gladia / Speechmatics)

Soniox, Gladia, and Speechmatics return the original transcript and the translation in the same request. There's no second AI step, so it's faster and you don't pay an AI provider on top of the transcription cost.

Set it up per mode under Edit Mode → Transcription:

  1. Pick Soniox, Gladia, or Speechmatics as the provider for the mode.
  2. Pick the source Language (or leave on auto-detect where supported).
  3. Pick the Translate to language. Leave it as Off if you want pure transcription.
Edit Mode screen with Soniox provider selected and a 'Translate to: Russian' picker highlighted

Every recording in that mode is now translated automatically into the target language as part of the transcription job. The original transcript is still saved alongside the translation, so you can revisit either side.

The available target languages depend on the provider:

  • Soniox - 60+ languages.
  • Gladia - 100+ languages.
  • Speechmatics - 50+ languages.

The Translate to picker only appears when you've selected one of these three providers - other providers don't offer translation at the transcription layer, so use the AI approach below instead.

AI Translation

For any other transcription provider (Groq, Deepgram, ElevenLabs, Cohere, Mistral, WhisperKit, Parakeet, etc.), translation runs as a second step using your AI provider and a translation preset.

Built-in Translation Presets

VivaDicta includes translation presets out of the box:

  • 🇺🇸 Translate to English
  • 🇷🇺 Translate to Russian
  • 🇪🇸 Translate to Spanish
  • 🇨🇳 Translate to Chinese
  • 🇫🇷 Translate to French
  • 🇩🇪 Translate to German
  • 🇧🇷 Translate to Portuguese
  • 🇯🇵 Translate to Japanese
  • 🇰🇷 Translate to Korean
  • 🇸🇦 Translate to Arabic
  • 🇮🇹 Translate to Italian

Each preset preserves the original meaning, tone, and structure of your speech - outputting only the translated text with no extra commentary.

Set Up Automatic AI Translation

Assign a translation preset as the default AI preset for a Viva Mode:

  1. Go to Settings and select a mode (or create a new one).
  2. Enable AI Processing and choose a translation preset as the default.
  3. Now every recording in that mode is automatically translated into the target language.

For example, set up a "Spanish Mode" - every recording you make in that mode gets automatically translated to Spanish.

Translate Existing Recordings

Already recorded something? Open it in the transcription detail view and generate a new variation with a translation preset. The original and all translations are saved side by side as separate variations.

Create Custom Translation Presets

Need a language that's not built in? Create a custom preset:

  1. Go to AI Presets in settings.
  2. Create a new preset with category Translate.
  3. Set the prompt to something like: "Translate the text into Japanese. Preserve the original meaning, tone, and structure. Output only the translated text."
  4. Give it a name and icon (e.g., 🇯🇵 Japanese).

You can create presets for any language your AI provider supports. For free AI translation, use Groq - the same free API key works for both transcription and AI processing.

Native vs. AI - Which Should I Use?

  • Use native when you're already on Soniox, Gladia, or Speechmatics. It's faster, costs less (no AI call), and the translation is purpose-built for transcribed speech.
  • Use AI when you're on a different transcription provider, when you need a language a native provider doesn't support, or when you want a more agentic transformation (translate and summarize, translate into a specific register, etc. via a custom preset).

Tips

  • Multiple translations - generate variations in different languages from the same recording in the detail view.
  • Combine with Assistant - with clipboard context enabled, you can copy text and say "translate this to French" using the Assistant preset.

See also: Live Translation, Presets & Variations, Viva Modes.

← AI AssistantPresets & Variations →