https://retellai.com/?utm_source=discord&utm_medium=seo logo
here problem with dialogue, transcript show 40 sec...
# ❓・ai-evy-support
l
@AI Evy here problem with dialogue, transcript show 40 sec, but conversation was 1 min 40 seconds call_74523679b2aa895c61c2703f2ea
a
Thanks for tagging me @lifo_fifo. I'll look through relevant sources and get back to you shortly.
Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.
### Answer I don't have specific documentation about transcript duration discrepancies in the available sources. However, there's a related community forum post about missing transcripts in webhook payloads [(1) ]() , which suggests transcript-related issues have been reported. For your specific issue where the transcript shows 40 seconds but the actual conversation was 1 minute 40 seconds, I recommend: 1. Check your agent's data_storage_setting configuration - it controls what gets stored [(2) ]() 2. Verify the max_call_duration_ms and end_call_after_silence_ms settings aren't cutting off the recording prematurely 3. Review the webhook payload to see if partial transcripts are being sent during the call via transcript_updated events If you'd like more detailed troubleshooting for this transcript duration mismatch, I'd suggest: Check out these potentially useful sources: [Create Voice Agent]() , [List Voice Agents]() You may also want to reach out through: - [Documentation]() - [Hire certified Retell partner]()
​ To learn more, see the sources I considered: [[1] Transkription wird nicht mitgesendet!](https://community.retellai.com/t/transkription-wird-nicht-mitgesendet/450) [[2] Create Voice Agent](https://docs.retellai.com/api-references/create-agent) ​
l
@AI Evy here are logs ] 2026-05-12 14:45:53.821 call_74523679b2aa895c61c2703f2ea info: Transitioning from node node-1768384883169 (name: date 1) to node node-1768410837658 (name: Conversation) 2026-05-12 14:45:54.527 call_74523679b2aa895c61c2703f2ea info: Latency for response_id: 2 - e2e: 3975 asr: 562 llm: 915 tts: 813 2026-05-12 14:46:01.277 call_74523679b2aa895c61c2703f2ea info: Transitioning from node node-1768410837658 (name: Conversation) to node node-1770625956702 (name: date 11) 2026-05-12 14:46:02.065 call_74523679b2aa895c61c2703f2ea info: Latency for response_id: 3 - e2e: 1553 asr: 563 llm: 2 tts: 784 2026-05-12 14:46:05.344 call_74523679b2aa895c61c2703f2ea info: Transitioning from node node-1770625956702 (name: date 11) to node node-1768410863918 (name: Logic Split Node) 2026-05-12 14:46:05.381 call_74523679b2aa895c61c2703f2ea info: Transitioning from node node-1768410863918 (name: Logic Split Node) to node node-1768384885855 (name: Extract Variables) 2026-05-12 14:46:05.381 call_74523679b2aa895c61c2703f2ea info: Calling tool: extract_dynamic_variables Arguments: 2026-05-12 14:46:06.321 call_74523679b2aa895c61c2703f2ea info: Received result for tool: extract_dynamic_variables_7d8b0d extract_dynamic_variables Result: {"year_of_birth":"1982"} 2026-05-12 14:46:06.421 call_74523679b2aa895c61c2703f2ea info: Transitioning from node node-1768384885855 (name: Extract Variables) to node node-1743142122526 (name: Introduction) 2026-05-12 14:47:08.223 call_74523679b2aa895c61c2703f2ea info: Ending call: call_74523679b2aa895c61c2703f2ea 2026-05-12 14:47:09.696 call_74523679b2aa895c61c2703f2ea info: Webhook triggered for call_ended 2026-05-12 14:47:10.173 call_74523679b2aa895c61c2703f2ea info: Webhook response received for call_ended 2026-05-12 14:47:14.744 call_74523679b2aa895c61c2703f2ea info: Webhook triggered for call_analyzed 2026-05-12 14:47:14.993 call_74523679b2aa895c61c2703f2ea info: Webhook response received for call_analyzed
a
Thanks for the follow-up, @lifo_fifo. I'll get back to you soon.
### Answer Based on your logs, the call ran from approximately 14:45:53 to 14:47:08, which is about 1 minute 15 seconds—closer to your observed 1 minute 40 seconds than the 40-second transcript suggests. To investigate the transcript discrepancy: 1. Retrieve the full call data via API - Use the Get Call API to fetch complete call details including
transcript
,
transcript_object
, and
duration_ms
[(1) ]() [(2) ]() 2. Check webhook payloads - Your logs show both
call_ended
and
call_analyzed
webhooks were triggered [(3) ]() . The
call_analyzed
webhook includes the complete
call_analysis
object with the full transcript [(3) ]() [(4) ]() 3. Verify transcript_object timestamps - The
transcript_object
field contains word-level timestamps that show exactly when each word was spoken [(1) ]() [(2) ]() . This can help identify if portions of the conversation are missing from the transcript 4. Review the recording - Compare the
recording_url
duration against the transcript to confirm the actual conversation length To access complete call data: You can retrieve the call using the call_id:
call_74523679b2aa895c61c2703f2ea
via the Get Call API, which returns all fields including
duration_ms
,
transcript
,
transcript_object
, and
recording_url
The webhook payloads for
call_ended
and
call_analyzed
events also contain this data—check what your webhook endpoint received [(3) ]() [(4) ](
.com/features/post-call-analysis-consumption) If the transcript is genuinely incomplete despite the full conversation occurring, this would be an issue requiring further investigation with Retell support.
​ To learn more, see the sources I considered: [[1] Update Call](https://docs.retellai.com/api-references/update-call) [[2] List Calls](https://docs.retellai.com/api-references/list-calls) [[3] Webhook Overview](https://docs.retellai.com/features/webhook-overview) [[4] Consume the analysis data](https://docs.retellai.com/features/post-call-analysis-consumption) ​
Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.