Noise Sensitivity Too High and Transcription Lag — Are There Configurable Parameters? #96
Description
Hi @pamelafox,
Thank you for your work on the RAG audio feature. I’ve been using it regularly and wanted to report a few concerns along with a question about potential configuration options.
Issues Observed:
High Noise Sensitivity
The system tends to pick up background or distant voices too easily, which affects transcription accuracy. It often captures unintended speech instead of focusing on the primary speaker.
Transcription and Response Lag
The feature has been noticeably slower recently, both in terms of generating transcriptions and responding to queries. This impacts usability, especially in real-time interactions.
Question:
Are there any parameters in the codebase that can be adjusted to reduce noise sensitivity or prioritize near-field audio? Specifically:
Can noise thresholds sensitivity be configured?
Is it possible to adjust microphone input sensitivity or filtering levels?
Are there any settings that affect the responsiveness or real-time performance of transcription and query handling?
If such parameters exist, guidance on where and how to modify them would be appreciated.