Issue with Twilio media-streams instead of Dialogflow Phone Gateway on Heroku (node.js)

[for-questions]

I decided to dispense with Dialogflow Phone Gateway.
I'm trying to use Twilio media-streams
I disabled Phone Gateway integration + removed webhook to Heroku backend in DF Fulfillment tab.
I have simple conversation flow with several follow-up intents with @sys.any parameters (accept any answer)
I have nothing inside Text response fields for all intents (logic requires to add answer using backend)
Basically I do not even need the audio response from Dialogflow
What I want is Intent's "Display name" value and "Query Text" value back (to know caller's answer)
Later I will send relevant response back to caller using @google-cloud/text-to-speech
I run it on local machine + ngrok. All works fine
BUT when I try it on Heroku it stuck on first intent (Welcome)
It continues to gather what I say and if I keep rather long silence and then say something - nothing happens
It does not shift to next follow-up intent
Only when I hang up -> new intent name becomes visible in logs
I have a feeling that the solution is simple but I really stuck
Will be very much obliged

HEROKU LOGS:
session: 3c0c74b6-508e-41d6-923e-9a7c2942d887
Captured call CA7c5203b5a18d858a20432ad1fcb2132d
First-Intent
Welcome
Detected intent First-Intent
not detected
session: 3c0c74b6-508e-41d6-923e-9a7c2942d887
Sending audio
not detected
Interrupted with "Nick"
Clearing...
not detected
Interrupted with "Nick"
not detected
Interrupted with " Nick"
not detected
Interrupted with " Nick"
MediaStream has finished
Disconnecting from Dialogflow
Second-Intent
Nick Nick
Detected intent Second-Intent
not detected
*
P.S. I found this old question
maybe it is about the same issue - https://stackoverflow.com/questions/5839435/do-we-need-a-static-ip-to-use-html5-websockets
if so - how to attach static IP to media stream?