# Release review · 1.0.0 · 2026-09-09

**Disposition: complete technical sample; not approved as a native Latin American Spanish production delivery.** No platform upload has been made.

## Completed on the original example

- Original English/Spanish product UI, ten state snapshots, twenty bilingual narration units and thirty UI/term pairs share `project.json` and version 1.0.0.
- Both 120-second H.264/AAC MP4s generated; narration is genuine local synthesis, not an empty placeholder. Caption-free MP4 preserves functional UI and truthful sample/voice disclosures.
- PCM16 mono 16 kHz narration is exactly 120 seconds; speech clips total 89.632 seconds, with reading pauses. Peak absolute sample 25,558; zero clipped samples. No audio acceleration or music.
- SRT/VTT cue times come from actual WAV boundaries. No word-level forced alignment or professionally verified reading-speed claim.
- Eighteen validation cases and thirteen state/syntax cases passed. Supplementary integrity tests check stale text/audio and packaging safeguards; see `integrity-qa.json`.
- Both complete MP4s decoded without reported errors. Five extracted frames were inspected for layout, accents and edge clipping. Windows newline spacing was corrected and output regenerated. See `media-qa.json`.
- No customer/server connection in the fictional UI. Its download is marked `real_customer_approval: false`. All data is synthetic.

## What these checks do not settle

1. **Latin American voice:** not closed. This artifact deliberately uses an es-ES Spain voice. The Spanish draft targets Latin America but has not had native regional review.
2. **Listening and linguistic acceptance:** no qualified human listened or proofread. Audio decoding/waveform checks do not prove naturalness, pronunciation or narration accuracy. The model is `x_low`, not a premium production-quality promise.
3. **Real customer product:** none supplied. Current Spanish UI, approved terms, facts, recording rights and final language/product sign-off must come from the actual buyer/authorized reviewer.
4. **Browser deployment:** Node state/syntax checks passed. A real-browser end-to-end interaction and mobile/accessibility audit were not performed in this asset environment; the integrator must perform them. The video is a same-source illustration, not evidence of recorded browser interaction.
5. **20-minute commercial operation:** this two-minute fixed demo does not prove turnaround, customer demand, reviewer supply or the old hypothetical unit economics. No sale or delivery to a customer occurred.

## Review acceptance sheet for a real release

| Gate | Evidence required | Current status |
|---|---|---|
| Source rights | Customer authorized footage, UI/environment and script | Not applicable to fictional sample; required for customer |
| Version freeze | Named product build, source/target UI dates and hashes | Sample 1.0.0 only |
| Glossary | <=50 approved term/UI pairs, named product approver | 30 author-draft sample pairs |
| Language | Native target-region editor checks script, captions and full audio | Not performed |
| Audio locale | Licensed suitable target voice, explicit region and listening pass | Spain fallback only |
| Visual match | Compare every action/label against supplied current target UI | Same-source fictional states only |
| Final delivery | Customer acceptance of corrected package, separate from internal technical QA | Not performed |

If any approved input changes, create a new release version, resynthesize affected clips, rerender and recheck. Two correction rounds in a future quotation would cover the agreed frozen source; a new product build or language is a new scope. Do not label technical hashes as a customer approval.

## Production input mode added in kit 1.1.0

The separate authorized-input CLI now accepts supplied target recordings, a reference video, approved WAV/script/timeline and <=50 terms for up to 20 minutes. An actual 36-second three-cut original-media fixture produced both MP4s, captions, WAV and a nine-file delivery ZIP. Eighteen additional schema/rights/hash/timeline tests passed; both 36-second videos fully decoded. See customer-mode/README.md and validation-tests.json. This closes the earlier fixed-renderer technical scope gap; it does not close native Latin American voice, linguistic approval or real customer evidence.