Create: Multilingual character
Last updated
Last updated
Moemate supports multilingual spoken conversations. And while there are multiple models available to have these talks, certain considerations should be taken into account in order to choose the right models for a specific language. We will explain here everything you need to know about setting up multilingual conversations with Moemate.
While setting up spoken multilingual conversations in Moemate there are various factors you need to consider. Not all LLMs are trained in multiple languages. Similarly not all voice models support voiced conversations in different languages. Select the models that support your native language to have the best experiences consistently.
First, select the relevant language(s) for the user and character based on the voice model. Follow these steps to select the language:
1-Edit character and go to Language Section
2-Select the character’s language from the available list. Note that only languages supported by the selected voice model appear here.
3-Select the user language.
Keep in mind that selecting the language here doesn’t hard code the language. Instead it just suggests to the LLM in our API calls that this might be the language spoken by the user in respective conversations.
As explained previously, the selectable languages that appear in the Basic Info section depend on the selected voice model. You can switch between voice models by going to the “Models” section in Edit/Create Character. There are two voice model providers available: ElevenLabs and Azure. ElevenLabs models are a PRO-only feature right now and support more languages with better supported voices. Azure has 50+ voice options. Scroll down in “Voices” in the “Models” section to see which voices are specialized in which languages. Note that not all voices support multilingual conversations so choose the right voice model for your character from the list.
Not all Large Language Models (LLMs) are trained in multiple languages. Most of the LLMs are fine tuned on data in English. So your LLM choice needs to be taken into consideration while setting up your character for multilingual conversations in your native languages as well. GPT and Claude are the LLM providers which support the most number of languages. Note that conversations in languages other than the ones listed above still might work. So, in conclusion, depending on what language you’d like your character to speak and understand, the most versatile selection of model providers for multilingual conversations would be ElevenLabs for voice and GPT/Claude for LLMs; however, there’s no hard and fast rule. Better to experiment and decide for yourself!
