> ## Documentation Index
> Fetch the complete documentation index at: https://docs.decktalk.app/llms.txt
> Use this file to discover all available pages before exploring further.

# Rebuild one section

> What DeckTalk caches and reuses after an edit, and how to re-synthesize, re-record, or re-assemble a single section with --only, --force, and the stage commands.

`decktalk build` is safe to run after every edit, because each stage reuses what has not
changed. This page says what is reused, and how to limit a build to one section when
you want to save time or credits.

## What an edit costs

| You changed                                             | What runs again                                                                                                                                                                            |
| ------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| A sentence in one section of `script.md`                | That section is synthesized and recorded again. The other sections' narration is cached, and their recordings are still valid, because cue times are relative to each section's own start. |
| A phrase in `cues.json`                                 | Nothing is synthesized. The section whose cues changed is recorded again.                                                                                                                  |
| A slide in `deck/index.html`                            | Nothing is synthesized. Every section that uses the page should be recorded again.                                                                                                         |
| `[voice]` settings, or the model                        | Every section is synthesized again, because the cache key includes the voice id, the model, the output format, and the voice settings.                                                     |
| A clip file, `hold_seconds`, `[transition]`, or `[mix]` | Only `assemble`.                                                                                                                                                                           |
| `extra_seconds`                                         | The section is recorded again, because the recording's length changed.                                                                                                                     |

The narration cache is `build/audio/manifest.json`. Each entry carries a hash of the
provider, the voice settings, and the exact text sent, and `narrate` skips a section whose
hash, file, and words file are all present. The recording stage has no cache of its own,
so `--only` is how a recording is spared.

## Limit a build to one section

```console theme={null}
decktalk build --only 3                  # narrate everything (cached), re-record section 3, assemble, verify
decktalk build --only 3 --only 4         # two sections; the flag repeats and does not take commas
decktalk build --only 3 --preset veryfast
```

With `--only`, the `record`, `measure`, and `check` stages touch only the sections named.
Narration still runs for every section, which costs nothing for the cached ones, and
`assemble` still cuts every section, because the final mp4 needs all of them. Name every
section whose text or page changed. A section left out keeps its old recording even if
its words moved.

## Run the stages by hand

Each stage is also a command, and the sequence below is what `build --only 3` does for
the recording.

```console theme={null}
decktalk narrate                         # cached sections are skipped
decktalk beats
decktalk record --only 3
decktalk measure --only 3
decktalk check --only 3
decktalk assemble
decktalk verify 3:3.1eq
```

After a page edit with no script change, `record --only N`, `measure --only N`, and
`assemble` are enough. After replacing a clip file, `assemble` alone is enough.

## Force a fresh synthesis

```console theme={null}
decktalk narrate --force                 # ignore the hash and synthesize every section
decktalk build --force
```

A forced run spends credits on every section. `decktalk narrate --dry-run` prints what
would be sent, section by section, and the character count follows from it.

## A silent build and a voiced build

The manifest records whether its entries are silent placeholders. A voiced build after a
silent one starts from an empty cache, and a silent build after a voiced one does the
same, so switching between them never mixes the two. Neither overwrites the other's
final mp4 until it finishes, and a timestamped copy of every finished video sits beside
`build/out/<name>.mp4`.

## Next

[How it works](/concepts/how-it-works#caching-and-cost) explains why a section's
recording stays valid when its neighbours change.