Hi everyone, I’m currently facing an issue with our VAPI-based call scenario. In our setup, callers are required to provide a 6-letter identifier by spelling it out using the NATO-style phonetic alphabet (e.g., “A as in Alfred, B as in Bernard,” etc.). This is the only method we’ve found that allows the callbot to reliably capture the identifier. However, we’re consistently running into a problem: the callbot interrupts the caller before they can finish spelling out the full identifier. It happens almost every time, regardless of the settings we’ve tried tweaking. We’ve experimented with all sorts of parameters but nothing seems to fix it. Has anyone encountered something similar or found a workaround for this? Thanks in advance!

Try increasing the

waitSeconds

in your

startSpeakingPlan

and enabling

smartEndpointing

to prevent the assistant from interrupting callers who are spelling out identifiers. For example:

json
"startSpeakingPlan": {
  "waitSeconds": 1.0,
  "smartEndpointingEnabled": true,
  "smartEndpointingPlan": { "provider": "livekit" }
},
"stopSpeakingPlan": {
  "numWords": 2,
  "voiceSeconds": 0.4,
  "backoffSeconds": 1
}

Test these settings and adjust as needed to ensure callers can finish spelling before the bot responds. [See more in the documentation.](https://docs.vapi.ai/customization/speech-configuration) Source: - [Speech configuration documentation](https://docs.vapi.ai/customization/speech-configuration)

Increase the speech timeout settings like waitSeconds and voiceSeconds, enable Smart Endpointing, and adjust transcription parameters to allow longer pauses, ensuring the bot waits for the full input before responding.

Try increasing the

waitSeconds

in your

startSpeakingPlan

and enable Smart Endpointing with the

livekit

provider to reduce interruptions when callers spell out identifiers. Here’s a sample configuration:

json
"startSpeakingPlan": {
  "waitSeconds": 1.0,
  "smartEndpointingEnabled": true,
  "smartEndpointingPlan": { "provider": "livekit" }
}

Test and adjust these values as needed to ensure callers can finish spelling before the bot responds. [See more in the documentation.](https://docs.vapi.ai/customization/speech-configuration) Source: - [Speech configuration documentation](https://docs.vapi.ai/customization/speech-configuration)

Hey! To help track down this issue, could you share: - The call ID - When exactly this happened (the timestamp) - What response you expected to get - What response you actually got instead This would really help us figure out what went wrong!

Thank you so much! By the way, do you happen to have any ideas on how to improve alphanumeric data collection — especially for letters? The only method we've found so far is to ask callers to provide a 6-letter identifier by spelling it out using a NATO-style phonetic alphabet (e.g., “A as in Alfred, B as in Bernard,” etc.). It's the only way we’ve been able to reliably match the ID with the customer database of our client. Unfortunately, it often fails — and since this is the very first step of the identification process, it’s been causing a lot of frustration on our client’s end. Most of our use cases are in French, and we’ve tried multiple different techniques without much success. Have you encountered something similar before? Or maybe seen a good solution shared somewhere on Discord or in the community that I could look into? Thanks a ton!!

Hey! To help track down this issue, could you share: - The call ID - When exactly this happened (the timestamp) - What response you expected to get - What response you actually got instead This would really help us figure out what went wrong!

Based on the documentation, increasing

waitSeconds

and enabling

smartEndpointing

(with

livekit

or

vapi

for non-English) in your

startSpeakingPlan

is recommended to reduce interruptions when callers spell out identifiers. You may need to experiment with these settings for French use cases. [See more in the documentation.](https://docs.vapi.ai/customization/speech-configuration) Quick question: Have you already tried switching the

smartEndpointingPlan

provider from

livekit

to

vapi

for your French calls, and if so, did it make any difference? Source: - [Speech configuration documentation](https://docs.vapi.ai/customization/speech-configuration)

@Trah1n can you please create a new ticket for data collection issue.