Transcripts drive captions, editing, and AI analysis. You can fetch full transcripts, chunked segments, and perform bulk corrections.

## Endpoints [#endpoints]

* `POST /v1/projects/{project_id}/transcribe`
* `GET /v1/projects/{project_id}/transcript`
* `GET /v1/projects/{project_id}/transcript/segments`
* `POST /v1/projects/{project_id}/transcript/corrections`

## Start a Transcription Job [#start-a-transcription-job]

```bash title="Terminal"
curl -X POST https://blitzreels.com/api/v1/projects/{project_id}/transcribe \
  -H "Authorization: Bearer $BLITZREELS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{ "media_id": "media-asset-uuid" }'
```

## Get Transcript [#get-transcript]

```bash title="Terminal"
curl "https://blitzreels.com/api/v1/projects/{project_id}/transcript?media_id={asset_id}" \
  -H "Authorization: Bearer $BLITZREELS_API_KEY"
```

## Get Transcript Segments [#get-transcript-segments]

Segments are useful for editing and AI‑assisted highlights.

```bash title="Terminal"
curl "https://blitzreels.com/api/v1/projects/{project_id}/transcript/segments?chunk_strategy=sentence" \
  -H "Authorization: Bearer $BLITZREELS_API_KEY"
```

## Bulk Corrections [#bulk-corrections]

Use bulk corrections for common typos or proper nouns. This updates both transcripts and caption words.

```bash title="Terminal"
curl -X POST https://blitzreels.com/api/v1/projects/{project_id}/transcript/corrections \
  -H "Authorization: Bearer $BLITZREELS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "media_asset_id": "media-asset-uuid",
    "replacements": [
      { "from": "Virgil", "to": "Virgile", "match_case": false, "match_whole_word": true }
    ],
    "phrase_replacements": [
      { "from_words": ["Cloud", "Code"], "to_words": ["Claude", "Code"], "match_case": false }
    ]
  }'
```

You can also use a `timeline_item_id` if you only have the clip reference.

`replacements` are single-token only. Use `phrase_replacements` for phrase fixes; v1 requires `from_words` and `to_words` to have the same token count so word timing stays stable.

## Tips [#tips]

* Poll job status or use webhooks before requesting transcripts.
* Use `chunk_strategy=time-window` for consistent segment durations.
* For single‑word corrections, use the Captions API.

## Related [#related]

* [Captions API](/docs/captions)
* [Jobs and Exports](/docs/jobs-exports)