Logseq/Entity/Podcast/Episode/GitP/A

• Ghost in the Patch Alpha Episode

  • In this garden, Ghost in the Patch Alpha Episode pages model recordings prepared as episodes of GitP/A.
  • The shared episode conventions come from Logseq/Entity/Podcast/Episode; this definition adds the session page and media preparation process.

  • Identity and sources

    • One episode lives at GitP/A/Session/YY/MM/DD-Day, using the recording date. Search that session path, the source recording filename, the music-making log, and distinctive title words before creating a page; an existing logical reference also counts as a page.
    • Start with the available recording and session notes. Follow existing source pointers to the project, music-making log, transcript, MIDI, and preset captures when present. A recording may come from a standalone instrument session or a DAW.
    • Missing optional material leaves the corresponding section absent. A project file, commentary, MIDI export, preset capture, or artwork is not a prerequisite for an episode.

  • Written for participants

    • An episode page is public. It is written first for GitP/Participants, the people who listen on the website or in a Podcast/Catcher, and second for performers who want to find a sound. GitP/Vibe sets the writing style for the opening description and the Ceremonial Annotations.
    • Each section earns its place, per My/Principle/Simplify. A fact appears once on the page, per My/Principle/Simplify/Don't Repeat Yourself DRY; the preset, MIDI and asset pages hold the details and the episode links to them.
    • Page links follow My/Pref/Markdown: plain wikilinks to garden pages. MIDI controller numbers stay off the page; a control is named by a link to its Microfreak page.

  • Episode page

    • Instances carry logseq-entity:: links to this definition and Logseq/Entity/Podcast/Episode, keeping the generic Episode membership explicit. Creator and recording-date properties follow the generic definition.
    • One H1 gives the episode title. Its children come in this order:
      • The recording, as a direct MP3 media embed, so the play button is the first thing a participant meets. The episode embeds the verified public media URL directly; the asset page holds the file metadata. Until Read the episode description when the recording comes first · Issue #138 · codekiln/gitpa lands, the Gitpa RSS build reads the show notes from the H1's first child, so published episodes keep the description first and the recording second.
      • The opening description: a few sentences that become the show notes in the RSS feed and the first words in a podcatcher. It is inviting, weird and curious, and carries no AI flange: no filler facts, hedges or bookkeeping that serve no part of the ceremony.
      • ## Ceremonial Annotations
      • ## Chapters
      • ## Ceremonial Artifacts, last.

  • Ceremonial Annotations

    • The official notes of the episode, in the voice of GitP/Vibe: what the ceremony was, what was summoned, and what a performer or participant might listen for.
    • Artwork appears here when available.

  • Chapters

    • Chapters follow the recording in time order. Each H3 is one preset's entry, written ### MM:SS enter [[Microfreak/Preset/<number> <name>]], and the blocks nested under it are about that preset. A passage with no preset change stays nested under the preset that is sounding.
    • Moments within a preset start with their own timestamp, at any depth: #### 03:07 secondo strato as a heading, or a plain bullet such as 05:15 - the bass comes in. Nesting makes chapters and sub-chapters. Times are measured against the episode recording, including any edits or offsets.
      • On the website, timestamps that start an H3–H6 heading seek the recording. Make timestamp-led bullets in chapter notes seek the recording · Issue #139 · codekiln/gitpa extends that to plain bullets at any depth.
    • Notes name their speaker:
      • What Person/codekiln wrote sits under a My Notes block.
      • What AI wrote sits under a GitP/Acolyte block, in the voice of the AI co-host.
      • Quoted notes stay as plain bullets, without Markdown block quotes.
    • Commentary from the recording, when present, sits beneath its chapter in a child block with collapsed:: true, each line starting with its timestamp.
    • The chapter at 00:00 enter Microfreak/Preset/016 Voidom in GitP/A/Session/26/10/29-Thu is hand-annotated and shows the target shape and voice.
    • Links to preset pages, guide sections and controls sit beneath the moment they explain. Confirm exact page names and aliases before adding links. Technical claims come from the recording, captured data, saved presets, or the linked documentation; unknown settings stay out.
    • MIDI analysis explains what a gesture did to the sound. Saved preset settings and recorded controller events describe different evidence and remain distinguishable. A useful negative fact stays, written as a link, such as the Microfreak/UI/Oscillator/Knob/Type never changing.
    • Logseq/Entity/Podcast/Episode/GitP/A/Layer names layers A, B and C and their audio and MIDI parts.

  • Ceremonial Artifacts

    • The archive for participants and performers who want to dig into how the sounds were drawn out of the instruments. Most readers skip it, so the H2 carries collapsed:: true and sits at the bottom of the page.
    • ### Recording gives the duration, the recording's asset page and a labeled download link, then how it was recorded: the DAW, tempo, instruments, layers, mix decisions and edits.
    • Other H3s follow when the material exists:
      • ### Presets lists each preset with a one-line description, its guide link and a labeled download link.
      • ### MIDI gives the labeled MIDI download and what the capture holds.
      • ### Set links the episode's Logseq/Entity/Ableton/Set page.
    • Pages published before this definition keep their ## Recording and Prep section until they are revised.

  • Publication selection

    • Logseq/Entity/Podcast/Episode/GitP/A/Frontmatter/public on the source episode in logseq-encode-garden controls whether the episode is selected for the downstream RSS feed and website.
    • Set public:: false for a draft or withdrawal and public:: true for publication. Missing values leave the episode unpublished after sync. Re-sync copies the source value even when the destination proxy previously had another value.
    • The changed selection reaches the feed and website after sync and the next publishing build and deployment.

  • Prepare and preserve media

    • Each source or derived file has its own page beneath the session at GitP/A/Session/YY/MM/DD-Day/Asset/<Purpose>/<format>, following Logseq/Entity/Asset. The purpose names the recording or version; the final segment is the lowercase file extension.
    • The page-derived filename replaces namespace separators with ___ and the final format segment with the extension. A local working copy lives at assets/.remote/<filename>.
    • Run the asset tasks from the source garden checkout, passing logical page names without wikilink brackets. Their reference pages hold access requirements and recovery details.
      • Asset/mise/Task/add — mise run asset:add -- <source-file> <asset-page> copies and DVC-tracks an original or separately prepared file.
      • Asset/mise/Task/fetch — mise run asset:fetch -- <asset-page> restores a tracked file needed for preparation.
      • Asset/mise/Task/convert — mise run asset:convert -- <wav-page> <mp3-page> registers a reproducible single-WAV-to-MP3 conversion.
      • Asset/mise/Task/push — mise run asset:push -- <asset-page> backs up each tracked source and prepared output.
    • A multitrack session needs a separate mix and export before single-WAV conversion. Mixing, commentary balance, and session-specific edits require their own preparation; the conversion task supplies no general multitrack mixing workflow.
    • Preserve the original recording. Commit the adjacent .dvc metadata and any conversion dvc.yaml and dvc.lock; working media stays ignored under assets/.remote/.
    • Public files use Logseq/Entity/Asset/B2. Create the asset page before uploading, then run GitP/mise/Task/media/upload with mise run gitpa:media:upload -- <prepared-file> <asset-page>.
    • Add the returned verified URL to the asset page and the episode's media or download link. The upload task checks that the public download matches the prepared file and has the expected media type. DVC backup and named public upload are separate operations.
    • Check public playback and seeking, artwork, and each supplied download using GitP/How To/Draft an Episode from Session Assets.

  • Optional session material

    • Existing preset captures can supply download links and documented settings. New device capture and renaming follow the instrument's existing workflow, including backup and protection of unsaved sounds.
    • Available MIDI can be exported for download after checking tempo and alignment against the episode recording. Automatic preset capture, preset renaming, and MIDI enrichment remain unfinished automation; the existing upload task accepts prepared files.

  • Publication

    • GitP/How To/Prepare Podcast Metadata describes publication preparation in Person/codekiln/GitHub/gitpa. Episode text and publication selection stay in this source garden; the publishing garden in that repository holds an exact-name proxy.
    • Preview Logseq/Entity/Proxy/Page/mise/Task/sync with explicit source and destination garden paths and the episode's logical page name; --apply writes the reviewed import. Sync the audio and other relevant asset pages separately, since direct media URLs do not make them embedded-page dependencies.
    • The source episode owns public::. Its destination proxy in the codekiln/gitpa repository stores podcast-guid:: and podcast-published-at::; the generic Episode definition preserves those release values on re-sync.
    • Run mise run rss:build, mise run rss:check, and mise run rss:test from Gitpa. Check the preview's recording playback, seeking, and downloads.
    • The publication PR gives the episode title, description, and direct MP3 link. The human listens once and decides whether to publish; merging the Gitpa PR lets CI publish the episode.