Back

Multilingual Voice Support

Multilingual Voice Support

GoSiteMe's voice system supports over 30 languages, enabling Alfred and your website's voice features to communicate with a global audience. This guide covers language selection, multilingual setup, and optimization tips.

Supported Languages

GoSiteMe voice supports the following language families, with multiple regional variants:

  • European: English (US, UK, AU), French (FR, CA), Spanish (ES, MX), German, Italian, Portuguese (BR, PT), Dutch, Polish, Swedish, Norwegian, Danish, Finnish, Greek, Czech, Romanian
  • Asian: Japanese, Korean, Mandarin Chinese (Simplified, Traditional), Hindi, Thai, Vietnamese, Indonesian, Malay
  • Middle Eastern: Arabic (MSA, Gulf), Turkish, Hebrew, Farsi
  • Other: Russian, Ukrainian

Setting Alfred's Language

  1. Go to Alfred → Settings → Language.
  2. Select the primary language for Alfred's responses.
  3. Enable Auto-Detect if you want Alfred to automatically respond in the same language the user speaks or types.
  4. Click Save.

Multi-Language Websites

If your website serves visitors in multiple languages, you can configure language-specific voice profiles:

  1. Create a voice profile for each language (see Creating Custom Voice Profiles).
  2. Under Alfred → Voice → Language Routing, map each language to its corresponding profile.
  3. When a visitor interacts in French, Alfred uses the French profile; when they switch to English, Alfred switches automatically.

Translation & Localization

Alfred can translate content in real time during voice interactions. Enable Live Translation under Alfred → Settings → Translation. This is useful for:

  • Customer support: Answer queries in the customer's language regardless of your team's language skills.
  • Content creation: Ask Alfred to draft content in one language and translate it to others.
  • Multilingual IVR: Route callers to language-appropriate voice menus automatically.

Optimizing Multilingual Voice Quality

  • Use native voice models for each language rather than a single English model speaking other languages.
  • Add pronunciation overrides for proper nouns that may differ across languages (e.g., city names).
  • Test each language with native speakers during setup to verify naturalness.
  • Adjust speed settings per language—some languages naturally have faster or slower speech rates.

Language Detection API

Use the Alfred API endpoint POST /alfred/v1/detect-language to identify the language of incoming text or audio before routing it to the appropriate voice profile. The API returns a confidence score for each detected language.

Was this answer helpful?