> For the complete documentation index, see [llms.txt](https://evidano.gitbook.io/ailyze-docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://evidano.gitbook.io/ailyze-docs/troubleshooting/transcription-issues.md).

# Transcription Issues

Fixes for transcription problems, from wrong speakers and locked editing to charges and shared links, plus tips for cleaner transcripts.

This page helps you fix problems with audio and video transcription and get cleaner results. Every control named here is described in [Transcription Settings](/ailyze-docs/reference/settings/transcription.md).

## "Accuracy is poor"

The transcriber does best with the right language setting and clean audio, so a wrong language or background noise lowers accuracy.

1. Pick the most specific language and accent. The **Language of Audio** list groups languages by **Best Accuracy**, **Better Accuracy**, and **Good Accuracy**, and suggests a better entry when one exists.
2. Add names, brands, and places under **Custom Dictionary**, and map spoken forms to written ones under **Custom Spelling/Formatting**.
3. For a recording that switches language, choose "2 or more languages in 1 audio (auto-detect)" at the top of the list.
4. After the run, open **Accuracy** in the editor and work through **Words to check**: the words the transcriber was least sure of.

## "Speakers are wrong or merged"

Evidano labels voices Speaker A, Speaker B, and so on, unless you ask for roles or names before submitting or with a re-run.

1. Before you submit, open **Speaker Labels** in **Optional Settings**, set **Number of Speakers** if you know it, and choose **Role** or **Name** under **Choose speaker identification**. These need a language in the Best or Better group.
2. After the run, click a speaker name to open **Rename speaker**, then choose **This turn only** or **Rename all**.
3. To relabel the whole transcript, open **More actions** › **Fix speaker identification**, pick **Role** or **Name**, and click **Re-identify speakers**. The wording does not change; you get an email when it is done.
4. If one paragraph holds two voices, click the first word of the second voice and choose **Start a new speaker here**, or use **Break into speaker segments** on the paragraph.

## "It's taking a long time"

Transcription runs in stages and longer recordings take longer, so a wait is normal.

1. Watch the status pill move from **Validating** to **Processing**, **Initial transcript**, **Enhancing with AI**, **Transcript ready**, and **AI reviewed**. The list refreshes itself while anything is running.
2. Wait for the email you get when the folder is done; you don't have to keep the page open.

## "Editing is locked"

**Edit text** stays greyed out until the transcript is ready.

1. Hover over the greyed button to read why, for example "Manual edit is available once the transcript is ready with AI review".
2. Wait until the file's status reads **Transcript ready** or **AI reviewed**, then click **Refresh** in the editor.
3. A shared link is view-only; ask the owner to add you with **Add as collaborator** (paid accounts only).

## "AI made edits I didn't want"

The AI review applies its corrections automatically and lists each one, so unwanted changes are easy to reverse.

1. Click **AI Edits** in the editor toolbar.
2. Read the **Before** and **After** of each change, and click **Undo** on any you disagree with.

## "Clicking a word doesn't play the audio"

A click on a word opens a small menu, so you can play, edit, clip, or split from that word.

1. Choose **Play from here** in the menu, or click the timestamp next to the speaker name.
2. Press Space to play or pause while not typing.

## "I can't find a PDF export"

Transcripts download as **DOCX**, **XLSX**, **SRT**, **VTT**, or **TXT** from **Download transcript**, so every format except PDF is built in.

1. Choose **SRT** or **VTT** for subtitles and **TXT** for plain text; tick **Include timestamps** for DOCX or TXT with times.
2. For a PDF, download the **DOCX** and save it as PDF from your word processor.

For the formats each workflow supports, see [Supported File Types](/ailyze-docs/reference/supported-file-types.md).

## "I was charged or hit the free limit"

Your free audio allowance is used first, and anything beyond it is charged to your wallet only after you confirm.

1. Read the blue banner above the files before you submit: it states how long the batch is, what is free, and what it will cost.
2. In the **Create project** dialog, check **Charge** and **Final balance** before you click **Create project**.
3. If your balance is short, **Transcribe … for free instead** trims the batch to the free time you have left and lets you pick the files to keep. It drops any translation.

For the rates, see [Plans, Limits & Credits](/ailyze-docs/reference/plans-limits-and-credits.md). Failed files show "You will not be charged."

## "A file was skipped"

Files with no audio are skipped automatically so the rest can go ahead.

1. Look for "The following file(s) do not contain audio and will be skipped:" after you submit.
2. Re-upload a version that contains audio if you still need it.

## "The person I shared with can't edit"

There are two ways to share, and only one lets the other person edit.

1. **Add as collaborator** gives a paid Evidano user edit access to the whole folder from their own account.
2. **Share through link** sends anyone a secure view-only link. They can read, search, download, and clip, but not edit.
3. To change how someone has access, remove them in the **Collaborators** list of the share dialog and share again the other way.

## Next steps

* [Transcribe Audio & Video](/ailyze-docs/prepare-research-data/transcribe-audio-and-video.md) — the full walkthrough.
* [Transcription Settings](/ailyze-docs/reference/settings/transcription.md) — every button and option.
* [Analyze Interview Transcripts](/ailyze-docs/common-research-workflows/analyze-interview-transcripts.md) — what to do with finished transcripts.