Phase 1 · Local model quality test

Bingo Audio Test v0.6.8

Whisper runs locally. v0.6.8 improves O-column recovery, confirms independent cross-window evidence, and suppresses duplicate audit extras. No beeps or fixed call slots.

Checking device…

1. Load the local speech model

WASM remains the recommended iPhone backend. Keep this tab open throughout the full audit.

Model not loaded.

2. Record a short benchmark clip

Use this for quick individual-call checks. Voice enhanced requests automatic gain and speech cleanup.

StatusIdle

3. Transcribe and evaluate the short clip

Enter each intended bingo call once. Repetition is evaluated separately.

Processing time
Audio length
Realtime factor
Expected recovered
Audio signal
Transcript quality

Raw Whisper transcript

No result yet.

Parsed call events

No calls parsed yet.

Elks repeat status

No repeated calls evaluated yet.

4. Automated 75-call audit

Use Natural Caller v3 on the laptop. The tracker now analyzes overlapping audio segments and aligns detected calls by order, so variable voice pacing and start delays do not break the audit.

Analysis methodSequence alignment

Direct-file baseline

Laptop-speaker room test

Tap Start room recording, then start Natural Caller v3.1 on the laptop. No beeps are required. Stop the phone recording after the caller says Done. The 12-minute safety limit is only a backup.

Audit statusReady

0/75 calls · 0%

Recovered
Confirmed twice
Provisional only
Missed
Analysis time
Matching

Current call transcript

No audit run yet.

Missed calls

No audit run yet.

Provisional-only calls

No audit run yet.

Unmatched detected calls

No audit run yet.

Results by letter

No audit run yet.

5. Parser-only test

Paste a Whisper transcript here to test the bingo parser without recording again.

6. Export diagnostics