I'm currently in Vietnam, Hanoi and wanted to see if I have any AI Voice Agent developers from vietnam in this community. If yes just drop a comment I would love to connect with some of you.

Hi everyone, I'm building a specialized voice assistant using **Pipecat Flows v0.0.22** and running into a frustrating issue with phone number collection that I can't seem to solve. ### The Stack - **Framework:** Pipecat Flows v0.0.22 (Python) - **STT:** Deepgram Nova-3 (Polish `pl`) - **TTS:** Cartesia (Polish voice) - **Transport:** Local WebRTC (browser-based, no telephony yet) ### The Problem When I dictate a 9-digit Polish phone number (e.g., "690807057"), the assistant receives partial fragments and processes them individually instead of waiting for the full number. For example, if I say "690... 807... 057" (with natural pauses), the bot splits it into: 1. "6" -> sent to LLM -> LLM complains "Received only 1 digit" 2. "980" -> sent to LLM -> LLM complains 3. "5" ... and so on. ### What I Have Tried I've gone through the documentation and tried several fixes, but the "defragmentation" issue persists. 1. **Deepgram Configuration (Current Setup):** I've configured the `LiveOptions` to handle phone numbers and utterance endings explicitly: ```python options = LiveOptions( model="nova-3", language="pl", smart_format=True, # Enabled numerals=True, # Enabled utterance_end_ms=1000, # Set to 1000ms to force waiting interim_results=True # Required for utterance_end_ms ) ``` *Result:* Even with `utterance_end_ms=1000`, Deepgram seems to finalize the results too early during the digit pauses. 2. **VAD Tuning:** - I tried increasing Pipecat's VAD `stop_secs` to `2.0s`. - *Result:* This caused massive latency (2s delay on every response) and didn't solve the valid STT fragmentation (Deepgram still finalized early). I've reverted to `0.5s` (and `0.2s` for barge-in) as `stop_secs=2.0s` is considered an anti-pattern for conversational flows. 3. **Prompt Engineering (Aggressive):** - I instructed the LLM to "call the function IMMEDIATELY with whatever fragments you have". - *Result:* This led to early failures where the LLM would call `capture_phone("6")`, which would fail validation (requires 9 digits), causing the bot to reject the input before the user finished speaking.

Am here to connect with like minded people and improve my knowledge.

Sharing builds, breakdowns, and lessons from working with AI in production. The stuff that actually matters when you're shipping to real businesses. 🔗 https://www.instagram.com/soumith_voiceai/

For those of you who want to see what a REAL CUSTOM CURATED voice agent sounds like and is capable of, please watch this video below. My team and I have been working on this agent for a while now but finally have it completed and in the hands of the customer. They are sooooo unbelievably happy with the outcome and are finally excited to no longer have to answer phone calls on the weekend. Probably the biggest win here is being able to spend quality time with the family again without getting interrupted. S/O to TrueWorks roofing for being absolutely AMAZING to work with throughout this project! https://www.loom.com/share/d98f1ec47c7541beac3b2a61247b3fda