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

# Uploading transcripts

> How to upload or paste interview transcripts onto a scorecard for AI scoring

You can attach a transcript to any scorecard by uploading a file or pasting text directly. Once attached, Waffle scores the candidate against every competency on the scorecard.

## Upload a file

<Steps>
  <Step title="Open a scorecard">
    Navigate to your scorecard library and open the scorecard you want to score against.
  </Step>

  <Step title="Find the transcript section">
    Scroll to the **Transcript** section on the scorecard detail page. You will see a drop zone with supported format labels.
  </Step>

  <Step title="Drag and drop your file">
    Drag a transcript file (VTT, SRT, JSON, or TXT) onto the drop zone — or click it to open a file picker. Files up to **10 MB** are accepted.
  </Step>

  <Step title="Preview and confirm">
    Waffle shows a preview of the parsed transcript with speaker labels, timestamps, and a segment count. Review the preview, then click **Attach transcript** to start scoring.
  </Step>
</Steps>

## Paste text

If you have transcript text in your clipboard (from a meeting tool, notes app, or email), you can paste it directly:

<Steps>
  <Step title="Open a scorecard">
    Navigate to your scorecard library and open the scorecard you want to score against.
  </Step>

  <Step title="Switch to the paste tab">
    In the **Transcript** section, click the text input option to open the paste box.
  </Step>

  <Step title="Paste your transcript">
    Paste the transcript text into the text area. Waffle auto-detects whether the text is VTT, SRT, or plain text.
  </Step>

  <Step title="Submit">
    Click **Attach transcript** to submit. If the pasted text matches a known format (VTT or SRT), Waffle extracts speaker labels and timestamps. For plain text, Waffle automatically detects common speaker-label patterns (such as "Interviewer:" or "Candidate:") and parses them into separate speaker turns.
  </Step>
</Steps>

<Tip>
  For best results, use a structured format like VTT or SRT. These preserve speaker labels and timestamps, which helps the AI identify who said what and produce more accurate competency scores.
</Tip>

## What happens after attaching

Once a transcript is attached:

1. **Normalization** — Waffle parses the transcript into a standard format with speaker roles, segments, and timestamps
2. **Deduplication** — if you upload the same transcript again, Waffle detects the duplicate and skips re-processing
3. **AI scoring** — the transcript is sent to the AI for competency scoring (this runs in the background)
4. **Status updates** — the transcript section shows a progress indicator while scoring is in progress, then displays results when complete

You can replace an existing transcript at any time by uploading or pasting a new one. The previous transcript is replaced and scoring runs again.

You can also **delete an attached transcript** to remove it and clear all associated scores from the scorecard. The delete option is available in the transcript section on the scorecard detail page.

## Scoring status

The transcript section displays the current scoring state:

* **Normalized** — transcript is parsed and scoring is in progress
* **Scored** — AI scoring is complete and results are visible on the scorecard
* **Failed** — scoring encountered an error (you can re-upload to retry)

<Info>
  Transcript scoring uses your plan's monthly token budget. If you exceed your budget, you will see a prompt to upgrade. See [Plans and pricing](/billing/plans) for details.
</Info>

## Supported file formats

| Format     | Extension | Speaker labels       | Timestamps           |
| ---------- | --------- | -------------------- | -------------------- |
| WebVTT     | `.vtt`    | Yes                  | Yes                  |
| SRT        | `.srt`    | Yes                  | Yes                  |
| JSON       | `.json`   | Depends on structure | Depends on structure |
| Plain text | `.txt`    | Auto-detected        | No                   |

## Frequently asked questions

<AccordionGroup>
  <Accordion title="What transcript formats work best?">
    VTT and SRT produce the best results because they include speaker labels and precise timestamps. Plain text works but may result in lower-confidence scores since the AI cannot distinguish between speakers.
  </Accordion>

  <Accordion title="Is there a file size limit?">
    Yes — files up to **10 MB** are accepted. Most interview transcripts are well under this limit.
  </Accordion>

  <Accordion title="Can I attach multiple transcripts to one scorecard?">
    Each scorecard supports one transcript at a time. Uploading or pasting a new transcript replaces the previous one and triggers a fresh scoring run.
  </Accordion>

  <Accordion title="What if my token budget is exceeded?">
    If your plan's monthly token budget is exhausted, transcript scoring is paused until the next billing cycle. You can [upgrade your plan](/billing/upgrading) for a higher token budget.
  </Accordion>
</AccordionGroup>