I’ve heard experts online recommend AGAINST having our AI voice agents attempt to say the caller’s name because of how often it mispronounces the name, which can annoy the caller. What is the current recommendation? This, by the way, would be for incoming calls, and the agent might ask the name of the caller. Should the agent repeat back like, “Hi Mary, how may I help you?” Common names like Mary would be easy, but there are plenty of very challenging names.
Hey @4mythings
You can use Echo Verification (repeats/spells back names for confirmation), Smart Matching (tolerates transcription variations like “Brandon” vs “Brendon”), and you can also use custom pronunciation dictionaries (IPA/CMU) for tricky words on supported 11Labs English voices.
Thank You
I’ve been debating this myself. I do find it to be a very awkward conversation when the AI Agent mispronounces the name that somebody just said. Users don’t seem to know they can correct it. I’m starting to lean towards not even asking for the name up front and only asking for the name if it’s required for things like bookings or when it’s it’s definitely needed. I think it’s better where it just answers says greeting and just says “how may I help you?” I think time is being wasted and annoying the caller by asking for a name upfront and trying to get it right.
I also prompt the agent to not not repeat back the caller’s name in case it didn’t get it right. I used to think it was good that the AI would start or end a responses with the caller’s name, but I think real people don’t talk like that and it’s a bit annoyting.
That’s just my opinion from what I’m seeing so far.
Awesome. I’ll check those out.
Do you have the AI repeat names in your incoming and outgoing agents? And have you found it to be effective enough that people aren’t annoyed by mispronounced names?