Session 17 Source Provenance

Source recording: https://recorder.google.com/bceaa1ac-b94d-4b0e-a732-706ebfec7e1b

Recorder title: Sep 15 at 6:45 PM. Displayed date: September 15, 2026. Duration: 10,226.005 seconds (2:50:26). Retrieved September 16.

The complete 340,300,005-byte M4A was retrieved from the recording’s observed playback resource using verified contiguous HTTP ranges. It contains two mono audio streams plus data streams. Stream 0 was copied without re-encoding, then compressed to 16 kHz mono Opus for Deepgram nova-3, latest diarization, utterances, paragraphs, smart formatting, numerals, punctuation, English, and wiki-derived keyterms. Deepgram request ID: 01a0ac83-93c1-77e0-8c42-b34578fc5ebb; 2,737 utterances. Primary transcript text is uncorrected ASR.

The Google transcript was extracted from all 905 page utterances, preserving their text, visible timestamps, and displayed speaker-label changes. It ends at 02:50:15. It is a page extraction, not a claimed native .txt download. A separate QA-only derivative supplies explicit [Speaker N] headings for the existing comparison script; the original extracted text remains unchanged.

Automated sequential alignment is only a triage aid: Google groups several Deepgram turns into a paragraph, diarization labels do not correspond across engines, and some passages include overlapping speech. Key claims were checked manually by timestamp against both sources. Neither transcript’s speaker numbers are accepted as a global player-to-character map. Raw audio is not included in the public site; recording URL and transcription artifacts preserve provenance.

Source: Session 17

Back to Home

Recording integrity

Complete downloaded M4A SHA-256: a73269b8efe23a93a3f1bb1f91cf7c09ab3701ec3b50fe33e4fc179dd30e8b90. Google page extraction has 905 utterances; primary Deepgram has 2,737 turns and duration 10,226.005 seconds.