> For the complete documentation index, see [llms.txt](https://docs.ccv.brown.edu/ai-tools/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.ccv.brown.edu/ai-tools/services/transcribe/viewing-editing-transcriptions.md).

# Viewing/Editing Transcriptions

{% hint style="info" %}
The transcription viewer is a full editor for the transcripts generated by Brown-operated AI models. You can review, correct, and de-identify (new in 2.0.0!) a transcription on-page before downloading.
{% endhint %}

### Opening a transcription

To open the transcription viewer/editor, click the name of an audio file on the "Viewing Job" page. To return to the whole job, click the **Back** (←) button at the top left of the page.

<figure><img src="https://3357215318-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FXaSdR1JjDGbvccnFNb3t%2Fuploads%2FpHT5j9QOHK4AfzvX0uF7%2Ftranscription-header.png?alt=media&amp;token=181df345-4a94-441d-bad2-5a8a767926d6" alt="Header of the transcription viewer showing the file name, duration, processing time, de-identification status, and the transcript rating."><figcaption></figcaption></figure>

The header shows:

* **File name** of the audio/video file.
* **Duration** of the file (clock icon).
* **Processing time** and how much faster than real time the file was transcribed (for example, `25s` / `20.8x`).
* **Audio deleted** (:mute: icon) appears if the audio portion of the job was deleted by you or by CCV (automatically after 1 week).
* **De-identified** (green :shield: icon) appears if a de-identified version of this transcription has been created. See [De-identifying a transcription](#de-identifying-a-transcription).
* **Rate transcript**: Please rate your transcriptions, especially the ones that didn't turn out well. Ratings show CCV where *Transcribe* works and doesn't so we can investigate problems and improve the service.
  1. Click a star next to **Rate transcript** in the header of the transcription page. The **Rate Transcript** dialog will open and the green check mark confirms your rating was saved.
  2. *(Optional)* In **Comments**, tell us what worked well and what didn't (up to 2,000 characters). For example, you could mention misheard words, wrong speaker labels, or missing sections.
  3. *(Optional, but very helpful)* Select **Allow the CCV team to review this transcript and its source file to investigate my rating**. This gives the Transcribe team your consent to access your transcription and source files (if not already deleted by you or automatically) to investigate any issues based on your feedback. You can revise your feedback and revoke your consent at any time by clicking the 'edit' :pencil2: button next to the star rating.
  4. Click **Submit** to send your comments and consent, or click **Close** if you only want to leave the star rating.

Two buttons on the right side of the header open consolidated help pages:

* **?** - Quick help for the Viewing Transcription page
* ⌨️ (keyboard) - A list of all keyboard shortcuts

The **⋮** (Options) button at the top right of the page opens a menu with the following actions:

* **Copy to Clipboard**: copies the transcript text.
* **Download transcript**: see [Downloading Transcriptions](https://docs.ccv.brown.edu/ai-tools/services/transcribe/downloading-transcriptions.md).
* **Download captions/subtitles**: see [Captions/Subtitles](https://docs.ccv.brown.edu/ai-tools/services/transcribe/downloading-transcriptions/subtitles.md). *This option is not available for transcriptions created with Gemini models.*
* **De-identify**: creates a de-identified version of the transcription. See [De-identifying a transcription](#de-identifying-a-transcription).

#### The toolbar

<figure><img src="https://3357215318-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FXaSdR1JjDGbvccnFNb3t%2Fuploads%2Fisphoqya8rwskrR1rYnD%2Fimage.png?alt=media&amp;token=1dee9814-2789-4c0b-9010-9f2c429f05c2" alt="A toolbar with three choices: &#x22;Edit Speakers&#x22;, &#x22;Group Sentences&#x22; and &#x22;De-identified&#x22; (shown only when de-identification has been run)."><figcaption></figcaption></figure>

The toolbar above the transcript has three buttons:

* **Edit speakers**: rename, add, or remove speakers. See [Add/remove speakers](#add-remove-speakers).
* **Group sentences**: switch between the default and grouped view modes. See [View modes](#view-modes).
* **De-identified**: switch between the original and the de-identified transcript. See [Toggling de-identification](#toggling-de-identification).

### Viewing transcriptions

#### View modes

**Default (sentence-by-sentence).** By default, *Transcribe* shows the transcription one sentence at a time as segments. Each segment shows the speaker with start and end timestamps and the transcript. Segments are color-coded by speaker - all segments from the same speaker have the same color.

**Grouped.** To combine consecutive sentences from the same speaker into one block, click **Group sentences** in the toolbar or press `Ctrl + Shift + G`. To go back to the default view, click the button or press the shortcut again.

{% hint style="warning" %}
You cannot edit, insert, delete, or select segments while sentences are grouped. \
Ungroup them first.
{% endhint %}

#### Navigating between transcription segments

Use the `Up (↑)` and `Down (↓)` arrow keys to move between segments. You can also use `Tab` to move through segments, or click a segment to focus it.

#### Add/remove speakers

Each speaker is identified to the best of the models ability. To manually tweak these results, begin with the **Edit speakers** dialog:

* Click **Edit speakers** in the toolbar, or&#x20;
* Select a transcription line and press `H` while a segment is focused.

In this dialog, press the :pencil2: pencil icon to the speaker on that line an aliases (for example, change "Speaker 1" to the interviewer's name) and press the :ballot\_box\_with\_check: check to complete. You can also add or remove speakers.

{% hint style="warning" %}
You can only remove a speaker if no transcription segments are assigned to that speaker.
{% endhint %}

### De-identifying a transcription

{% hint style="warning" %}
**De-identification is experimental.** It can miss identifying information or redact text that isn't identifying. Always review the de-identified transcript before you share it. If you find a problem, use the **Please help us improve it** link on the transcription page to report it.
{% endhint %}

You can de-identify a transcription from the transcription page. De-identification attempts to identify names, places, phone numbers, and other personally identifiable information (PII) in the transcript, and replaces each one with a placeholder label in square brackets:

| Original                                                                                | De-identified                                                                                                                                                                                                                                                   |
| --------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| …And they did, too, for Josiah Carberry.                                                | …And they did, too, for <mark style="color:$success;">\[first\_name]</mark> <mark style="color:$success;">\[last\_name]</mark>.                                                                                                                                 |
| Carberry is the president and CEO of the nonprofit economic think tank Psycho Ceramics. | <mark style="color:$success;">\[last\_name]</mark> is the <mark style="color:$success;">\[occupation]</mark> and <mark style="color:$success;">\[occupation]</mark> of the nonprofit economic think tank <mark style="color:$success;">\[company\_name]</mark>. |

#### Creating a de-identified version

{% hint style="info" %}
[Captions](https://docs.ccv.brown.edu/ai-tools/services/transcribe/downloading-transcriptions/subtitles.md) and word-level JSON cannot be de-identified.
{% endhint %}

1. Open the transcription.
2. Click the **⋮** (Options) button at the top right of the page.
3. Select **De-identify**.
4. Review the **De-identify this transcript?** dialog, then click **De-identify** to confirm. To go back without de-identifying, click **Cancel**.

{% hint style="danger" %}

* Each transcript can be de-identified only **once**.
* **Make all your edits first,** as later edits to the original are not carried over (or vice versa).
  {% endhint %}

When the de-identified version is ready, the header shows a green **De-identified** badge, and you can view it with the newly established **De-identified** button in the toolbar.

#### Toggling de-identification

Click **De-identified** in the toolbar to switch between the two versions:

* **Highlighted (green):** the de-identified transcript is shown, and redacted information appears as green bracketed placeholders.
* **Not highlighted:** the original transcript is shown.

{% hint style="info" %}
The original transcript is not deleted when you de-identify. You can switch back to it at any time with the **De-identified** button.
{% endhint %}

### Editing transcriptions

You can reassign speakers, edit timestamps, and change the text of each segment.

{% hint style="warning" %}
You can edit segments only when sentences are not grouped.&#x20;
{% endhint %}

When you hover over a segment or focus it with the keyboard, four buttons appear on that segment:

<figure><img src="https://3357215318-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FXaSdR1JjDGbvccnFNb3t%2Fuploads%2FxfdgN21DH76daXxsLjZN%2Fimage.png?alt=media&#x26;token=99340935-2035-4ddc-874f-f3aef8203194" alt="Editing controls shown for a transcription segment"><figcaption></figcaption></figure>

#### Entering edit mode

In edit mode, you can change the timestamps and the text of the segment, and reassign the segment to a different speaker.

To enter edit mode, do one of the following:

1. Mouse over the segment and click the **Edit** button (pen icon).
2. Select the segment press `J`.
3. Click the timestamp of the segment.

#### Entering insert mode

Insert mode works like edit mode, except that your text is added as a new segment below the current one.

To enter insert mode, do one of the following:

1. Click the **Insert** button (+ with 3 adjacent lines icon, next to edit).
2. Select the segment and press `K`.

#### Saving or canceling an edit

When you finish editing or inserting a segment, click **Save** (`Shift + Enter`) to save your changes. To discard them, click **Cancel** (`Esc`).

#### Deleting a segment

To delete a segment, click the **Delete** button (trash can icon) or with the segment selected, press `Delete`, `Backspace`, or `L`.

{% hint style="danger" %}
Deleting a segment cannot be undone.
{% endhint %}

### Bulk editing

You can select multiple segments, then merge or delete them at the same time.

#### Selecting multiple segments

To select a series of segments, click its checkbox (right-most icon) or press `Space` while the segment is focused. You can only select segments when sentences are not grouped.

While at least one segment is selected, you cannot edit individual segments. The per-segment edit, insert, and delete buttons are hidden, and the bulk action buttons below are enabled instead.

<figure><img src="https://3357215318-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FXaSdR1JjDGbvccnFNb3t%2Fuploads%2FQhOO6AZdBcz9gPGiaiv2%2Fimage.png?alt=media&amp;token=7cfb4c20-9e31-4522-b6d4-c3608307c9d0" alt="A series of icons reading &#x22;3 selected&#x22;, &#x22;merge&#x22;, and &#x22;delete&#x22;." width="375"><figcaption></figcaption></figure>

#### Bulk actions

* **Deselect all** (`Ctrl + Shift + E`)
* **Delete selected** (`Ctrl + Shift + R`)
* **Merge selected** (`Ctrl + Shift + W`): available when you have selected at least two consecutive segments from the same speaker.

### Audio playback

Playing the audio back while you review the transcription helps you check its accuracy. If the audio file is still available, an audio player appears at the bottom of the page, and playback keyboard shortcuts are enabled.

{% hint style="warning" %}
Audio playback is available for up to 7 days after a job is completed. After that, the audio file is deleted from *Transcribe*. We recommend reviewing your transcription within this window.
{% endhint %}

* **Play/Pause (`Ctrl + Shift + S`)** <img src="https://3357215318-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FXaSdR1JjDGbvccnFNb3t%2Fuploads%2FtM6qR8H1doaP1DWUjYU1%2Fimage.png?alt=media&#x26;token=be90da10-b21c-4865-b293-d640f30935fb" alt="Play or pause button" data-size="line">\
  Starts or pauses playback. The segment that is currently playing is underlined.
* **Back 5 seconds (`Ctrl + Shift + A`)** <img src="https://3357215318-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FXaSdR1JjDGbvccnFNb3t%2Fuploads%2Fhkh4ZgCZSszgDYz57wCj%2Fimage.png?alt=media&#x26;token=76419402-1044-4956-b721-ee39acdc59a3" alt="Back 5 seconds button" data-size="line">\
  Moves the playback back 5 seconds.
* **Forward 5 seconds (`Ctrl + Shift + F`)** <img src="https://3357215318-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FXaSdR1JjDGbvccnFNb3t%2Fuploads%2FR9e6xK84zLlv4UxJn38M%2Fimage.png?alt=media&#x26;token=e9e8f857-23a7-495d-9dac-4f6e6c9ae7c4" alt="Forward 5 seconds button" data-size="line">\
  Moves the playback forward 5 seconds.
* **Auto-scroll (`Ctrl + Shift + D`)** <img src="https://3357215318-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FXaSdR1JjDGbvccnFNb3t%2Fuploads%2FiFFwoPD55z8vJobGYCSQ%2Fimage.png?alt=media&#x26;token=6ef33f69-b23a-4565-a0dc-7293b012eef7" alt="Auto-scroll button" data-size="line">\
  When auto-scroll is on, the page scrolls to follow the audio as it plays.

You can also play a single segment. Click the round button to the left of the speaker's name, or press `Enter` while the segment is focused.


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://docs.ccv.brown.edu/ai-tools/services/transcribe/viewing-editing-transcriptions.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.
