I have created a flow using nodes and the agent doesnt acknowledge me answering until the 2nd time I answer. A good example would be this call ID: 0f78d69c-541b-44ec-a381-f5b93ff9bd80

Hi there! 👋 I understand how frustrating it can be when your assistant doesn't recognize your input on the first attempt. I don't have access to the specific details of your call ID (0f78d69c-541b-44ec-a381-f5b93ff9bd80), but I can definitely help with the issue of your Vapi assistant requiring two attempts to recognize your answers in your node-based flow. ## Potential Causes There are several reasons why this might be happening: 1. **Endpointing Configuration Issues**: According to the [documentation I found](https://discord.com/channels/1211482211119796234/1211483291191083018/threads/1338497725917827196), the endpointing settings might need adjustment. The

smartEndpointingEnabled

setting and

waitSeconds

parameters can affect how quickly the assistant recognizes when you've finished speaking. 2. **Transcriber Settings**: The transcriber configuration might need tweaking. Some users have found success by [changing from Deepgram to Whisper using Talkscribe or Azure](https://discord.com/channels/1211482211119796234/1211483291191083018/threads/1343612796717371393) as their transcriber. 3. **Turn-Taking Behavior**: The way [VAPI's audio-text model manages turn-taking](https://discord.com/channels/1211482211119796234/1211483291191083018/threads/1336989548357353524) could be affecting input recognition, especially with short responses. ## Troubleshooting Steps Here are some steps you can take to resolve this issue: 1. **Adjust Endpointing Configuration**: Try updating your configuration with these settings:

json
   "start_speaking_plan": {
     "waitSeconds": 0.8,
     "smartEndpointingEnabled": true
   }

And for the transcriber:

json
   "transcriber": {
     "model": "nova-2",
     "language": "en",
     "provider": "deepgram",
     "endpointing": 300
   }

When looking at the transcriber setting for my agent, I cant see Whisper like you suggested: Transcriber Settings: The transcriber configuration might need tweaking. Some users have found success by changing from Deepgram to Whisper using Talkscribe or Azure as their transcriber.

Hi there! 👋 I see you're looking for Whisper in your transcriber settings. I understand that you can't see Whisper directly listed as a transcriber option. That's because in Vapi, Whisper is actually accessed through a provider called Talkscriber. Talkscriber uses the Whisper model for its transcriptions. According to the [Vapi documentation](https://docs.vapi.ai/customization/multilingual), Vapi supports speech-to-text through providers like Deepgram (nova-family models) and Talkscriber (which uses the Whisper model). When you want to use Whisper, you need to select Talkscriber as your provider. To use Whisper via Talkscriber, you would configure your transcriber settings like this:

json
"transcriber": {
  "provider": "talkscriber",
  "language": "en"
}

Other transcriber options available in Vapi include: - Deepgram - Gladia - Assembly AI - Azure Speech Transcriber As mentioned in my previous message, some users have reported [better results](https://discord.com/channels/1211482211119796234/1211483291191083018/threads/1343612796717371393) by switching from Deepgram to either Talkscriber (for Whisper) or Azure as their transcriber when facing input recognition issues. Is there anything else you'd like to know about configuring these transcriber options? Sources: - [Vapi documentation on multilingual support](https://docs.vapi.ai/customization/multilingual) - [Vapi documentation on Talkscriber](https://docs.vapi.ai/providers/transcriber/talkscriber) - [Discord Thread on Transcriber Options](https://discord.com/channels/1211482211119796234/1211483291191083018/threads/1343612796717371393) - [Vapi documentation on Assembly AI](https://docs.vapi.ai/providers/transcriber/assembly-ai)

Looking into it...

Thanks for the reply. Where do I update this, as im using the visual assistant builder. Thanks

You need to visit the VAPI dashboard. Select your assistant and then select the advanced tab. Scroll down to the section named 'Start Speaking Plan' where you will find the options to enable these values or input these values through the dashboard.