3. On-Device Processing
All core features of XSpeak run entirely on your device (unless you choose to use an OpenAI-compatible model):
- Transcription of meetings is performed in real-time using on-device audio processing.
- AI analysis (summarization and insights) is powered by an on-device language model.
- If you create a voice profile, it is stored locally in the form of anonymized embeddings and remains on the device as long as the app is installed. When you label other speakers, their voice embeddings are also learned and stored locally on your device.
All processing and analysis are performed locally by default. However, if you choose to use an OpenAI-compatible model, your data may be sent to the external endpoint you configure for processing. In such cases, XSpeak does not guarantee the privacy or security of that data.