Source files
Everything behind the findings, as plain files. The full set lives in the GitHub repo: github.com/sharshi/lein-practice.
Reports (Markdown)
- research-chabad-trop.md — the full research report: taamim system, Chabad melody, text data, existing tools, voice fork, recommended path.
- motif-table-validated.md — the validated motif table: 23 motifs measured against Slavin's recordings, verdicts, what changed, what remains uncertain.
- taamim-voice-scope-2026-11-17.md — scoping notes: Torah mode v1, the two architectures, the Vayetzei first-aliyah proof.
- chabad-trop-research.md — the Chabad-melody deep dive behind the motif transcription.
Validation scripts and data (validate/)
- trope.py — trope parser: Sefaria text to per-word trope labels, with position disambiguation and legarmeh handling.
- phrases.py — phrase builder (verses to phrases by disjunctive) and the working-model degree table.
- measure.py — harmonic-product pitch measurer (octave-robust fundamental estimation and stable-note extraction).
- segment.py — RMS pause segmentation of a recording.
- bereishis-text.json, gen1.json, target-verses.json — the Sefaria texts used.
- whisper_bereishis_300.json, whisper_gap_5_30.json — whisper word-level alignments (see the reliability caveats in the validated table).
- plot_crepe_0-12.png — CREPE pitch track plot, first 12 seconds of Bereishit.
The remaining scripts in validate/ are the exploration and analysis helpers used during validation. The audio recordings (Slavin's MP3s) are Chabad.org's and are not included.