BURKI STANDARD PIPELINE BACKGROUND-AUDIO WORKSHEET
Original free resource | Prepared October 24, 2026

This is a setup and acceptance plan, not a completed listening test. No clip was
uploaded, previewed or played in a call for this resource. Provider, platform,
telephone and storage services can have separate costs.

1. PURPOSE AND CHANGE OWNERSHIP
Business task:
Why add this particular background clip?
Assistant / organization reference (no secrets):
Change owner:
Nonproduction assistant or approved change window:
Current publication state:
If legacy live settings, how is a save's production effect controlled?
Who may approve configuration and later test spending?

This concerns added output ambience. Caller-input noise filtering and transfer
hold audio are separate settings and separate acceptance questions.

2. PERMITTED SOURCE RECORD
Clip name / version:
Owner and permission record:
Permitted use and any restrictions:
Private conversations or identifiable customer material excluded? Evidence:
Local original retained at:
File format / size / duration:
File hash, if available:

Choose one source:
[ ] Owned MP3/WAV upload, within the editor's 10 MB limit
[ ] Eligible public audio-file URL under our control

Upload receipt / selected filename, if an upload is later performed:
URL source and access/expiry owner, if selected (no private tokens here):
Source changed since approval? YES / NO / UNKNOWN
Load/decode evidence, if available:
Unresolved source issue / owner:

A stored upload or typed URL is not evidence of actual route playback.
Do not paste storage credentials or private signed links into this worksheet.

3. EXACT CONFIGURATION RECORD
Editor path: Voice > Speech settings > Advanced voice stack > Background sounds
Conversation engine: Standard pipeline, verified from current choices
Background sounds enabled: YES / NO
Selected source / version:
Positive volume setting:
Loop continuously: ON / OFF
Saved at / saved evidence:
Publication/readiness reviewed separately:

Use the enabled switch off to disable. Do not use a slider value of zero as a
reliable UI mute procedure; the current editor has a default-volume fallback.
GPT Live is not the configuration path for this background-audio feature.

4. ORIGINAL FICTIONAL CANDIDATE
Business: Alder Studio
Task: workshop reception enquiries
Source: a studio-owned 20-second room-tone WAV with no speech
Proposed volume: 15% (not a calibrated level or a universal recommendation)
Proposed loop: ON
Status: NOT TESTED

The source is a fictional planning record. No audio file is supplied or claimed
to have been recorded for this template.

5. FREEZE THE COMPARISON
Baseline label / configuration evidence:
Candidate label / clip hash / positive volume / loop:
Instructions, voice and model choices held constant:
Intended route (browser and telephone recorded separately):
Listening device and environment:
Reviewer:
Number of attempts planned:
Other variables that changed and why:
Authorized allowance / spending owner:

Acceptance rule decided before listening:
- Important business status and identifier must remain intelligible.
- No inappropriate sound, distracting repeat join or unexpected clip ending.
- Reviewer can hear the caller and assistant in the intended conversation.
- Actual observations, including failed attempts, are retained.

6. ORIGINAL UNPERFORMED FIXTURES
S1 - business status
The workshop enquiry is a request. A team member will confirm the available
dates. What would you like the team to review?
Important distinction: request versus confirmed date.

S2 - correction and identifier
The reference is B7, not D7. Would you like me to repeat it?
Important distinction: B and D remain distinguishable; correction preserved.

S3 - longer reply and quiet pause
To help the studio review your enquiry, please tell me the workshop name and
your preferred day. A preferred day is not yet reserved. The team will explain
the available options before confirming anything. Which workshop do you mean?
Important distinction: speech remains clear throughout, then the caller has
space to answer; review the clip during both speech and the pause.

7. OBSERVATION ROW - COPY FOR EACH FIXTURE / ROUTE / ATTEMPT
Fixture / attempt / route:
Baseline observation and evidence:
Candidate observation and evidence:
Important words clear? YES / NO / UNRESOLVED
Correction/identifier clear? YES / NO / UNRESOLVED
Clip appropriate and free of private material? Evidence:
Loop join or clip ending distracting? Evidence:
Different behavior during assistant speech / idle pause:
Actual source/load failure, if any:
Accepted? YES / NO / UNRESOLVED; reason:

8. SPECIFIC CORRECTION AND RECHECK
Observed problem:
Single change to try (lower positive volume, revise permitted clip, disable):
What stayed constant:
Same fixture and route to reassess:
Observation / supporting receipt:
Remaining unknown:
Decision owner / date:

If the source fails to load, investigate access, size, selection and decoding;
raising volume does not repair a missing source. If caller recognition is the
problem, assess the input path separately rather than crediting output ambience.

9. RELEASE RECORD FOR THE BUSINESS'S OWN REVIEW
Settings saved / publication state evidence:
Intended route readiness and costs reviewed:
Actual listening result and accepted scope:
Browser-only limitations:
Telephone-specific result, if separately authorized:
Recording contents inspected, if used as evidence:
Follow-up owner / next review trigger:
Decision: KEEP OFF / REVISE / READY FOR SCOPED ROUTE REVIEW / ACCEPTED FOR SCOPE

Free template does not mean free production calls or an approved test budget.
Companion guide:
https://burki.dev/blog/burki-standard-pipeline-background-sounds
Media mechanism reference:
https://docs.livekit.io/agents/multimodality/audio/background-audio/