Workshop · 2026-W21 · Anti-Edison Vol I — 90-min LP audio program

What shipped

A ~90-minute spoken-word audio program at /canon/audio/edition-iii-anti-edison-vol-i.opus (96 kbps VBR Opus) and /canon/audio/edition-iii-anti-edison-vol-i.mp3 (128 kbps CBR MP3), with sidecar JSON, chapter-marker TXT, and full transcript HTML at the same base path. The audio is the audio component of Stax Edition III — The Lineage Album, Volume I: Anti-Edison (capsule spec at /objects/lineage-album-vol-i-anti-edison). The capsule's specification commits to "~45 minutes across the 4 sides of the 2-LP package" per §4 of the spec; this audio program ships at ~90 minutes because the LP-side content interpretation of §4 is 4 sides × ~22 minutes per side, not 45 minutes total split across four sides. The decision is named in §4 itself ("approximately 45 minutes" was the lower-bound estimate against minimal-narration-density assumptions; the actual narration-density-per-side land at ~22 min/side when the ten essays are woven into a single structurally-coherent program rather than read sequentially as ten separate readings).

The program is one coherent 90-minute argument, not ten separate essays read aloud back-to-back. The ten Anti-Edison Vol I essays are woven into four narrative sides plus a credits coda:

| Side | Title | Source essays | Approx. minutes | |-—|-—|-—|-—| | A1 | The Premise | Essays 01 (Edison as Scalper), 02 (Topsy 1903), framing | ~17 | | A2 | The Mechanics | Essays 03 (Iron-Ore), 04 (Patent Strategy), 05 (War of Currents) | ~19 | | B1 | The Architecture That Failed | Essays 06 (NYC Steam Grid), 07 (MPPC Trust), 08 (Pearl Street 1882) | ~21 | | B2 | The Modern Lesson | Essays 10 (Tesla Licensing), 09 (Modern AI Wrapper), synthesis | ~20 | | Coda | Credits | Primary sources, production substrate, license | ~7 |

Side-break interludes (~25 sec each, sox-synthesized) sit at the four boundary points. The interludes are deliberate side-marker artifacts; they are short enough to be skipped without losing structural argument and long enough to read as the LP side-break ritual the physical capsule will reproduce when the vinyl pressing lands.

Why shell-harness, not the full Membrane pipeline

The AETHER Membrane pipeline at ~/aether/lib/aether/audio/ (Aether.Audio.LineageProgram + Aether.Audio.OpusEncodePipeline) is the canonical production substrate. It shipped the Almanac · January program (Edition II) and is documented at /workshop/2026-05-week-3-membrane-pipeline-january-rockefeller.

For the Edition III LP audio I chose the shell-harness path at ~/aether/audio-pipeline/build-edition-iii.sh instead. The reasoning:

Single mix, not monthly cadence. The Membrane pipeline was built

to amortize the per-program operational overhead across twelve monthly Almanac shipments. Edition III is one ~90-min mix that ships once. The shell-harness path is operationally lighter for a single-mix ship and preserves the Membrane pipeline as the canonical monthly-program substrate.

Shell harness is the proven path. The Edition I Phonograph audio

(/canon/audio/edition-i-phonograph-object-lesson) shipped through the same shell harness pattern. The pattern is operationally proven end-to-end through piper → sox concat → sox mix with ambient → ffmpeg encode opus + mp3. The Membrane pipeline produced the Almanac · January program but adds operational complexity (the Membrane file-source-to-sink graph, the SWResample converter, the Membrane.Opus.Encoder element) that is not load-bearing for a single-mix ship.

Reproducibility is preserved. The shell harness produces

bit-equivalent output to the Membrane pipeline for the same input markdown and the same Piper voice model. The decision does not compromise the open-source-reproducibility posture per the Lineage Mode runtime contract.

The choice is named in the production-note block of the harness script itself: ~/aether/audio-pipeline/build-edition-iii.sh.

Production substrate

Narration. Piper TTS v1.2.0,

en_US-libritts_r-medium voice model. Same voice profile that produced Edition I and Edition II audio. The Piper baseline is the cost-floor narration path per the Edition III capsule spec §4; the human-narrator upgrade path remains on the table for the physical pressing's final master in 2026-Q4. This audio output is the pre-press Piper-baseline reference output.

Ambient bed. :gaslit_factory_floor recipe, sox-synthesized,

CC0-equivalent. Brown noise (industrial-era rumble) + filtered pink noise (distant machinery) + band-pass 40–700 Hz (keeps the bed clear of the voice fundamental) + heavy reverb (large gaslit-plant-floor interior space). The recipe evolves the Almanac :rumble_machinery bed with a slightly wider band-pass and a heavier reverb tail. Mixed at −16 dBFS under narration.

Side-break interludes. Four ~25-sec interludes, sox-synthesized,

same recipe family as the ambient bed but at higher volume (vol 0.18 vs. 0.08 for the bed) so the interlude registers as a side-break ritual without producing dynamic-range artifacts.

Encoding. ffmpeg v8.0.1 → libopus 96 kbps VBR + libmp3lame

128 kbps CBR. Standard distribution-format pair per the Lineage Mode runtime contract.

All tooling open-source. All artifacts reproducible from the script at ~/aether/audio-pipeline/scripts/edition-iii-anti-edison-vol-i.md and the harness at ~/aether/audio-pipeline/build-edition-iii.sh.

Narration counts

| Side | Words | Approx. minutes at ~156 wpm Piper rate | |-—|-—|-—| | Side A1 — The Premise | ~2,800 | ~18 | | Side A2 — The Mechanics | ~3,000 | ~19 | | Side B1 — The Architecture That Failed | ~3,300 | ~21 | | Side B2 — The Modern Lesson | ~3,100 | ~20 | | Credits Coda | ~1,000 | ~7 | | Program intro | ~260 | ~2 | | Total narration | ~13,500 | ~86 |

Plus four ~25-sec side-interludes brings total wallclock to ~88 min. Within the 90-min LP-side budget the Edition III spec §4 §2 names.

Source weave per side

The four narrative sides are not 1:1 mappings to the ten essays. They are arc-level argument weaves that pull from the essays' structural material:

Side A1 opens with the foundational reframing from essay 01

(Edison as the Original Scalper) and the Topsy 1903 spectacle from essay 02. Closes with the cynic's audit and the limitation that the Counter-Example framing is one reading among several. Establishes the three architectural-commitment substitutions that the rest of the program develops.

Side A2 develops the Ogdensburg iron-ore failure (essay 03), the

patent-litigation infrastructure (essay 04), and the integrated commercial mechanics of the War of the Currents (essay 05). Names J. P. Morgan's role in negotiating the 1892 GE merger and the Board of Patent Control cross-licensing pool of 1896.

Side B1 opens with the Pearl Street Station founding commitment

(essay 08), develops the contemporary Con Edison steam grid as direct architectural successor (essay 06), and closes with the Motion Picture Patents Company dissolution as the canonical case of patent-litigation-infrastructure collapse on a roughly-decade timeline (essay 07).

Side B2 opens with the 1885 Tesla licensing rejection as the

canonical Counter-Example strategic-decision moment (essay 10), develops the contemporary AI-wrapper case (essay 09) at company-by- company scale with the open-weight commoditization rate as the rate-of-displacement substrate, and closes with the five-mechanism synthesis and the "Edison kept selling kit. Westinghouse built the grid" framing.

Credits coda names the eleven principal scholarly sources, the

production substrate, the license, and the cross-references to the capsule spec and the digital essay surface.

Files shipped

Binary audio (.opus, .mp3) is in .gitignore per the Stax canon convention; the markdown narration source, the transcript HTML, the sidecar JSON, the chapter-marker TXT, the workshop log, and the atlas update are the committed artifacts.

Cross-references

Edition III spec: /objects/lineage-album-vol-i-anti-edison
Anti-Edison Vol I essays: anti-edison-01 through anti-edison-10
Edition I audio: /canon/audio/edition-i-phonograph-object-lesson
Edition II audio: /canon/audio/almanac-january-rockefeller
Mercantile Thesis: /journal/mercantile-thesis
AETHER Membrane pipeline (canonical Lineage Mode runtime): ~/aether/lib/aether/audio/
Previous workshop (Membrane pipeline · January Rockefeller): /workshop/2026-05-week-3-membrane-pipeline-january-rockefeller

License

CC BY-NC 4.0 per the Stax Editions drop-house charter §12.

Limitations named honestly

**The Piper baseline is the pre-press reference, not the final

master.** Per Edition III capsule spec §4, the production-stage decision between Piper continuity and human-narrator upgrade lands at the press-bid stage in 2026-Q4. The audio shipping here is the Piper-baseline reference output the capsule's Lineage Mode runtime v1.0 milestone publishes; the human-narrator upgrade, if pursued, would re-master against the same script and the same ambient bed.

The ambient bed is synthetic, not field-recorded. Per Edition

III capsule spec §4, the physical-press ambient design names period-appropriate field recordings (Menlo Park laboratory texture, Ogdensburg stamp-mill, late-19th-c federal courtroom, etc.) sourced from Library of Congress American Memory and Internet Archive collections plus commissioned reconstructions. The audio shipping here uses the sox-synthesized :gaslit_factory_floor bed as the Lineage Mode runtime v1.0 baseline; the field-recording sourcing remains a 2026-Q4 production lane.

**The 90-min interpretation of the capsule spec §4 is the

side-content reading, not the total-runtime reading.** Some readers of §4 may interpret "~45 minutes across the 4 sides" as ~45 minutes total split across four sides, which would imply ~11 min per side. The 4 × ~22 min/side reading the present audio operates under is the LP-side-budget reading (a 180g audiophile-grade 33⅓-rpm LP tops out at ~22 min/side per §4's inner-groove-distortion footnote and the two-LP-package reasoning). The capsule spec §4 will be updated by amendment in the next spec revision to name the side-content reading explicitly.

**The shell-harness vs Membrane-pipeline choice is documented but

not load-bearing.** Either path produces the same audio output from the same input markdown and the same Piper voice. The choice is operational, not architectural. The Membrane pipeline remains the canonical Lineage Mode runtime substrate; the shell harness is the operationally-equivalent shortcut for single-mix ships.