I’m currently planning to develop a simultaneous interpretation app for my mother, who loves to travel. Since she often encounters unstable internet connections abroad, I’m looking for a way to make the app functional even when offline.
I have envisioned the following workflow for the app:
1. AI-based Search and Local DB Configuration (On-device SLM)
The app will embed a lightweight model on the device. For a specific destination (e.g., Vietnam), the AI will pre-fetch data, such as local cuisine, ingredients, activities, regional information, ethnic and linguistic characteristics, and historical facts, to be stored in a local database. This data can be downloaded or updated whenever a Wi-Fi connection is available.
2. DeepL Voice API Integration and Offline Capabilities
My goal is to implement seamless simultaneous interpretation by integrating this local database with DeepL Voice. Specifically, I have two questions here:
Connectivity: Is an active internet connection mandatory for all real-time voice translation requests using the DeepL Voice API?
Offline Caching: If a persistent connection is required, is it technically feasible to pre-generate and store an "audio corpus" or a "translation cache" (based on the pre-fetched data) during the initial setup to support offline or semi-offline functionality?
I want to ensure the interpretation remains fluid even in areas with poor connectivity. I would greatly appreciate any insights or technical advice you could share.
I would greatly appreciate any insights or technical advice you could share. Thank you in advance for your help! 😊