Poor audio from older phones is common in healthcare use cases. Instead of writing them off, you can add a preprocessing layer to denoise and enhance audio before transcription. I’ve seen good results routing calls through Twilio or similar, applying noise suppression/VAD, then sending cleaned audio to STT engines like Deepgram or your existing Vapi setup. It won’t be perfect, but it usually improves transcript accuracy significantly for low-quality callers. I can help you @Jimmy