plain-purple
plain-purple•2y ago

Twilio IVR Latency

I want to add in filler utterances while VF processes a response to the user. Can anyone give a suggestion on the best way to implement?
5 Replies
Tico | Voiceflow Assistant
To implement filler utterances in Twilio IVR while your Voiceflow processes a response, you can consider the following approach: - Utilize the Dialog Manager API to manage the conversation flow. When you send a request to advance the conversation, you can include filler utterances as part of the response to the user. This can be done by customizing the response payload to include the filler utterance while Voiceflow processes the user's input or prepares the next response. - Customize the response traces in your Voiceflow project. Each trace, produced by a particular block in your project, can be tailored to include messages or visual cues as filler content. This way, you can ensure that the user receives immediate feedback, even if it's a simple acknowledgment, while the system is processing. Remember, the key is to maintain engagement with the user during any processing delays, ensuring a smooth and natural conversation flow. For more detailed guidance on customizing responses and using the Dialog Manager API, you can refer to our documentation: - Dialog Manager API Reference - Managing Intents and Utterances within the Assistant
plain-purple
plain-purpleOP•2y ago
Thanks Tico - you're a great bot, but I think I need some human help on this one :-}
conscious-sapphire
conscious-sapphire•2y ago
@NiKo | Voiceflow We see delays primarily at blocks listening for intents and of course at AI blocks. Is it possible to: 1) Add timers into the code that will initiate filler sentence(s) after some amount of time? Or 2) Somehow on the canvas add in utterances between a Users message and intent listening/AI blocks? Trying to figure out best way to improve the user experience.
plain-purple
plain-purpleOP•2y ago
Would really appreciate ANY guidance on this issue! 🙏
NiKo | Voiceflow
NiKo | Voiceflow•2y ago
We are working on supporting streamed responses to improve that, in the meantime you can use a text step with your message followed by an empty listen step (Capture/Choice) with a no reply set to 1 second. If you’re creating flows that tend to generate a long process/response time, you can also use a custom action with “stop on action” enabled to handle that part in your integration (or using an extension if you use the Chat Widget).

Did you find this page helpful?