Overview
WeTalkPal uses third-party artificial intelligence services to power conversational practice, speech recognition, text-to-speech, pronunciation feedback, grammar correction, and fluency scoring. This page clearly identifies the AI models and providers behind those features.
AI Models by Feature
The following table describes which providers and models are used for each AI-powered feature in the Service:
| Feature | Provider | Model / Service |
|---|---|---|
| AI conversation, grammar correction, pronunciation feedback, and fluency scoring | DeepSeek (via OpenRouter) | deepseek/deepseek-chat |
| Speech-to-text (voice transcription) | OpenAI (via OpenRouter) | openai/whisper-1 |
| Text-to-speech (AI partner voice) | Microsoft | Edge Neural TTS (e.g. en-US-AnaNeural) |
AI Infrastructure Gateway
Language model and speech-to-text requests are routed through OpenRouter (openrouter.ai), a third-party API gateway that connects WeTalkPal to the underlying model providers listed above. OpenRouter acts as an intermediary and does not use your data for its own model training.
How Your Data Is Processed
When you use speaking practice features, your voice recordings may be sent to the speech-to-text provider for transcription. Your messages and transcriptions may be sent to the language model provider to generate conversational responses, corrections, and fluency assessments. Text responses may be sent to the text-to-speech provider to produce spoken audio. We use these services solely to deliver the learning experience described on our website and in our app. For more information on data handling, see our Privacy Policy.
Changes to AI Models
We may update the AI models or providers we use to improve quality, reliability, or cost efficiency. When we make material changes to the models listed on this page, we will update this disclosure and revise the \"Last updated\" date. Continued use of the Service after such changes constitutes acceptance of the updated disclosure.
AI Limitations
AI-generated responses, pronunciation feedback, and fluency scores are provided for language learning purposes only. They may contain errors and should not be relied upon for professional, legal, medical, or other critical decisions. WeTalkPal does not guarantee specific learning outcomes.
Contact
If you have questions about our use of AI models or providers, contact us at [email protected].