https://vapi.ai/ logo
Join Discord
Powered by
Assistant staying quiet after couple of questions
# support
e
See image fora an example, for some reason my assistant takes a while to think after a couple of questions. Basically stops answering. The callId is: c7afb01e-4187-4424-a863-bd5aa54e6fb8 https://cdn.discordapp.com/attachments/1257838587261489174/1257838588121452604/image.png?ex=6685dcf8&is=66848b78&hm=dc1c08d1106b3c35cf86a5259e5c1066272df6a493bb1fa26e7734c40b35e1dc&
v
message has been deleted
@Esrever Please ask support for more info on this.
You can continue this conversation by mentioning me in the message.
e
@Vapi please give me more info
v
@Esrever I'm here to help with any questions or issues you might have with VAPI. However, regarding the specific issue with the assistant taking a while to respond or stopping after a couple of questions, I would need more details to provide a precise solution. If you're experiencing performance issues, it could be related to various factors such as network latency, server load, or a bug in the system. To address this, you might want to: 1. Check your internet connection to ensure it's stable and fast. 2. Ensure that there are no ongoing service outages or maintenance that could affect response times. 3. If the issue persists, gather details such as the exact time the problem occurs, steps to reproduce the issue, and any error messages that appear. With this information, you can reach out to the support team for a more in-depth investigation. If you'd like, I can assist you in closing this ticket if you feel your immediate concerns have been addressed.
e
The model sends an end token but does not read the answer outloud
Here is callId from a call I just tried right now: dbf61500-e246-4831-b535-1c7faf9f5d24
The "Assistant speech started" checkpoint does not start after the "Model sent end token" checkpoint.
@User can you look into this for me
s
@Shubham Bajaj can you look into it?
e
This is very urgent! Please assist me with this problem
@Shubham Bajaj have you had the time to check on this? My clients are waiting so it would be much appreciated if you could help me
s
Hey @Esrever I am in meet, will check right after the meeting.
e
Alright thanks
s
hey @Esrever can i get the another call id with same issue.
e
3d8e73dd-d0cb-4d84-b1f9-08482331d287
I have many more if you need
Just let me know
s
send 2-3 more.
Hey @Esrever send 1 or 2 more.
e
41562ea7-22fa-40f4-ad51-dbbab7c4ff98
3f50ecb2-e854-4318-8463-caa706cdc6d7
s
Hey for this TTS azure output voice output timeout.
e
d8671173-1279-4db5-a48a-05c24ca24bb1
98afa9ba-4dee-4c16-a3ef-579439aa3eba
All of these calls had the same problem
Something I can fix on my end?
861a62f9-b48e-4e90-8f6f-1ce5aa0b90a7
Just called again to test^
Same problem
When I ask that question "what about the bistro" the assistant does not read the answer outloud but I can see that the AI generates an answer
when checking the logz
logs*
s
Hey your using over 10 knowledge base files which does generate response but till then it reaches the TTS timesout.
e
Okay so my knowledg base has too ma h files?
s
yes
First try using it w/o knowledge base and then use it 4-5 files and then later use it with all files. If same happens at any of the stage ping me we will connect and resolve it.
e
Okay, I will try that. What is the max files I can use with Vapi KB?
I have all of the files connected to a Voiceflow agent (text to text based) and it works perfectly.
Therefore I used the same files in Vapi, should I use fewer filesÐ
And why does they agent stay "silent"?
Because I can see that the LLM generates the correct answer when I look at the backend in Vapi
Please let me know before I test ^
s
Hey I will get back to you on files limit.
Agent didn't stayed it generated the response but the voice part timeout and if the user input to agent output time crosses >5secs, we abort the agent response.
e
Okay so lowering the knowledge base files will fix that?
s
Not sure we have to figure it out, check DM.
e
Will do
PreviousNext
Powered by