TWILIO HD AUDIO / DOWNSAMPLED: CALL-LEG AUDIT WORKSHEET Reviewed October 14, 2026. Free resource; calling and diagnostics may be billed. Guide: https://burki.dev/blog/twilio-voice-insights-hd-downsampled This is a blank review plan with a fictional example, not an executed call audit. Burki does not currently ingest Voice Insights tags. Use only authorized records. 1. WRITE THE BUSINESS QUESTION What went wrong, in the caller's task? ______________________________________ Direction: incoming recognition / outgoing speech / other: __________________ Critical value or correction: ______________________________________________ Expected behavior: _________________________________________________________ Caller complaint or observed event, with source: ____________________________ Immediate staff action needed: _____________________________________________ Do not let an unresolved technical hypothesis delay the appropriate follow-up. 2. IDENTIFY THE ACTUAL PROVIDER LEG Twilio account/subaccount reference (no credentials): _______________________ Actual carrier-leg CA Call SID, verified from: ______________________________ Other parent/child leg references, when relevant: ___________________________ Call time, timezone and direction: __________________________________________ Burki conversation/application identifier: _________________________________ Routing or provider evidence connecting the records: ________________________ Correlation state: verified / candidate / unverified: ________________________ Do not assume an LK, ivr_lk_ or browser identifier equals a Twilio CA SID. Do not join records solely because their time or destination is similar. 3. INSPECT EXISTING AUTHORIZED SUMMARY Open Twilio Monitor -> Insights -> Voice -> Calls; inspect the actual call. Summary retrieval time: ____________________________________________________ Processing state: partial / complete / unavailable: _________________________ Observed edge and stream direction: ________________________________________ Negotiated codec in that record: ____________________________________________ Tags actually present: _____________________________________________________ Relevant jitter/packet-loss or other observation: ___________________________ Source screenshot/receipt kept privately: __________________________________ If partial or unavailable, retain a pending/unknown result, not "no issue." API access requires the provider's account prerequisites; no paid feature was enabled and no provider request was executed to produce this worksheet. 4. WORKED FICTIONAL CASE Caller requests the Willow Annex. Initial transcript says Willow Avenue. Caller corrects it: "Annex, not Avenue." Suppose a properly matched carrier summary shows HD Audio and Downsampled. Transport finding: record the tag observation under the provider's rule. Recognition finding: check whether the correction and negation survived. Application finding: check whether the final captured destination is Annex. Business finding: verify any staff follow-up or resolution separately. Cause: unresolved until evidence supports a specific path or application issue. Do not infer that conversion caused the name error or that a new model fixes it. 5. RECORD YOUR OWN SEPARATE OBSERVATIONS Transport observation and exact source: ____________________________________ Available transcript around the critical correction: ________________________ Final captured value and where it was persisted: ____________________________ Assistant acknowledged the correction? _____________________________________ Available authorized audio and recording point: _____________________________ Evidence of caller understanding, if any: ___________________________________ Action receipt or staff delivery evidence: _________________________________ Resolution evidence, if any: ________________________________________________ Unknowns that cannot be settled by a codec tag: ______________________________ 6. FAILURE CHECKS (RESULT FIELDS INTENTIONALLY BLANK) Case Observation Decision / owner Wrong provider call leg ______________________ _______________________ Unverified Burki correlation ______________________ _______________________ Partial summary ______________________ _______________________ API access unavailable ______________________ _______________________ Correct transcript/stale data ______________________ _______________________ Missing caller correction ______________________ _______________________ Audio from another point ______________________ _______________________ No action or resolution proof ______________________ _______________________ 7. BOUND THE COST AND NEXT STEP Existing diagnostics available? ____________________________________________ Additional collection actually needed? _____________________________________ Account/tier/currency quote and date: _______________________________________ US public Advanced Features example, October 14: USD 0.0024 per minute. This is separate from phone, number, AI and platform charges; actual terms vary. Owner of any proposed paid activation or controlled test: ___________________ Current task: read-only review / waiting for evidence / approved test: _______ Assigned technical owner and bounded next action: ___________________________ Caller request owner and follow-up state: ___________________________________ Recheck date and unresolved question: _______________________________________ Primary references: https://www.twilio.com/en-us/changelog/voice-insights-now-tags-hd-audio-calls https://www.twilio.com/docs/voice/voice-insights/api/call/details-call-tags https://www.twilio.com/docs/voice/voice-insights/api/call/understanding-edges https://www.twilio.com/docs/voice/voice-insights/api/call/call-summary-resource https://www.twilio.com/en-us/voice/pricing/us