Final transcription output omitting words/sentence...
# support
j
Call ID: 019b8ff8-33ea-7ffb-b1af-8a8b27c2a2e8 Timestamp: Jan 05, 13:02:49.910 Issue: Words/sentences missing in final transcript but present in Deepgram logs Expected: All spoken content should appear in the final transcript
v
Action Step: Navigate to your Vapi dashboard and go to
Observe > Call Logs
, then search for your Call ID (
019b8ff8-33ea-7ffb-b1af-8a8b27c2a2e8
). Review the full call transcript and error messages to pinpoint where words or sentences were omitted. Tip: Include the call ID and timestamp when reaching out to support for faster troubleshooting. Example:
Copy code
# In Vapi dashboard:
Observe > Call Logs > Search: 019b8ff8-33ea-7ffb-b1af-8a8b27c2a2e8
For more details, see the [Vapi debugging guide](https://docs.vapi.ai/debugging). Source: - [Debugging voice agents](https://docs.vapi.ai/debugging)
j
This entire sentence is missing from the final transcription:
Copy code
{
  "id": "log-134",
  "level": 30,
  "time": 1767647004578,
  "timestamp": 1767647004578000000,
  "severityText": "INFO",
  "severityNumber": 9,
  "body": "Deepgram final transcript",
  "attributes": {
    "category": "transcriber",
    "callId": "019b8ff8-33ea-7ffb-b1af-8a8b27c2a2e8",
    "orgId": "9e42a66f-1188-4531-a6c9-9b17d4ec13fa",
    "provider": "deepgram",
    "channel": "assistant",
    "transcript": "Anything else you'd like to add to your order?",
    "confidence": 1,
    "numWords": 9,
    "duration": 1680
  }
}
@User
@Vapi Support Bot
v
Thank you for sharing the missing transcript details. To help diagnose the issue, could you confirm if you see any error messages or warnings in the
Observe > Call Logs
section for Call ID
019b8ff8-33ea-7ffb-b1af-8a8b27c2a2e8
, especially around the timestamp when the sentence was omitted? This will help determine if the omission is due to a processing or integration error. For further troubleshooting steps, you can refer to the [Vapi debugging guide](https://docs.vapi.ai/debugging). Source: - [Debugging voice agents](https://docs.vapi.ai/debugging)
j
No, i see no errors or warnings in the logs @Vapi Support Bot
v
Thank you for confirming that there are no errors or warnings in the logs for your call. Based on the information available, the sentence from the Deepgram transcript—"Anything else you'd like to add to your order?"—was present in the Deepgram logs but did not appear in the final transcription output. Since there are no error messages or warnings in the Vapi call logs, this suggests the omission may not be due to a logged processing or integration error. At this point, without additional error details or system messages, I am unable to determine the exact cause of the missing sentence in the final transcript. If you have access to any further logs or system outputs that might provide more context, reviewing those could help identify the root of the issue. If you have any other details or questions, feel free to share them!
j
@Vapi @User can you please help us out here?
c
Hi jay, Thanks — I dug through the docs and related threads. Next steps I recommend you run now so we can escalate with full evidence: 1. In the Vapi dashboard, open Observe → Call Logs and search Call ID 019b8ff8-33ea-7ffb-b1af-8a8b27c2a2e8. 2. Export or screenshot: - The full Call Logs "Transcripts" tab (final transcript shown to users). - The Deepgram final transcript log entry you pasted (or the surrounding Deepgram logs showing the timestamp). - Any API/Webhook logs near timestamp Jan 05, 13:02:49.910 (Request/response entries for model/tool calls). 3. Confirm whether the Deepgram transcript appears in artifact.messages or only in the raw provider log. (You can check the call artifact in the same Call Logs page.) 4. If available, attach the mono recording URL from the call artifact (artifact.recording / artifact.mono.assistantUrl) or a short clip around that timestamp. Why: the evidence you provided shows Deepgram produced the sentence, but we need to verify whether Vapi received it into artifact.messages (transcript pipeline) or whether it was lost between the transcriber and the artifact/aggregation step. If you can upload the screenshots/exports here I’ll review and, if needed, open an internal escaltion referencing: - Call ID above - Timestamp Jan 05, 13:02:49.910 - Deepgram log snippet you pasted Documentation: https://docs.vapi.ai/debugging.
j
nvm, i built a script to generate correct transcription on our end using raw logs