Hi team, could anyone confirm if Assistable has plans to integrate Eleven Lab's dynamic language detection and/or Speech Synthesis Markup Language (SSML)?
Below is a bottle neck that I ran into where an agent set to a language seems to only accept the respond in the same said language. This means that when it asks me for my name in Chinese, and I say Frank, it wouldn't capture it. But we're in North America and we use English names too...
ChatGPT response to this below:
- If you set the agent’s default language to Chinese, the platform routes all speech-to-text recognition into Chinese mode → which means English replies get misrecognized or dropped.
- If you set it to English, then it won’t properly parse Chinese turns.
- Right now, Assistable likely doesn’t support dynamic auto language detection at the STT (speech recognition) layer — even if your TTS (ElevenLabs) can handle multilingual synthesis.
In other words:
- The assistant can speak both English and Chinese, but
- It can’t hear both unless the STT recognizer supports bilingual switching.
✅ Workarounds
- Confirm STT engine settings in Assistable
- Force English as default recognition, but add a manual switch trigger in your prompt:
- Split into two agents (if Assistable doesn’t support multilingual STT yet):