Fon isn’t supported well by most Vapi STT/TTS tools yet, so there’s no real plug-and-play option for it. What usually works is using Whisper for transcription and pairing it with a more flexible TTS like Coqui XTTS, especially if you can feed it some local voice data. At your scale, even a small batch of Fon recordings can make a noticeable difference if you fine-tune a bit.
Do you already have any real call recordings in Fon, and how strict is your response time requirement? @LeCorellian