Custom dictionaries
Our Custom Dictionary functionality biases the speech recognition algorithms used by Amazon Transcribe and Speechmatics towards a predefined set of terms, improving transcription accuracy for content containing:
- Product names
- Technical jargon
- People’s names
- Brand-specific terminology
This is particularly useful when you know the vocabulary that's likely to appear in your video.
Custom dictionaries are language-aware rather than locale-specific. This means a single dictionary can be reused across regional variants of the same language, such as English (UK), English (US) and English (Global).
Creating a custom dictionary
Before using the Custom dictionary feature, you’ll need to create one:
- Open the Custom dictionaries page from the sidebar.
- Click New custom dictionary.
- Give it a name, select the language for your dictionary and provide a description if needed.
- Click Create & add terms.
- Add your terms (we recommend fewer than 1,000 entries) using one of the following methods:
- Add terms manually: type them manually or paste them in bulk (from a text editor or spreadsheet)
- Paste list of terms: paste the terms separated by new lines
- Import from CSV: choose or drag and drop a CSV file
- Generate terms from a source: extract terms from a document (up to 25,000 characters at a time)
- Click Submit or Update to save the terms to your dictionary.
Note: Large custom dictionaries can reduce transcription quality, although they do not affect transcription latency. If your dictionary contains a large number of entries, CaptionHub will display a warning.
The Sounds like field is optional and can be used when a term's spelling does not clearly indicate its pronunciation. This is particularly useful for names, product names, abbreviations or terms such as Porsche 911 or .NET.
You can also add an optional Note to record any additional information about the term.
CaptionHub will automatically suggest when a pronunciation hint may be helpful, although you can also refer to the Custom Dictionary examples for guidance.
Adding terms directly from the caption editor
You can also add words to a Custom Dictionary while editing your original captions.
- While editing a caption, highlight and right click the word you want to add.
- Select Add to Custom dictionary.
- Choose the target dictionary and edit the Text and Sounds like fields if required.
- Click Submit.
The selected word is immediately added to the dictionary and will be available for future transcriptions, allowing you to improve recognition without leaving the editor.
Using a custom dictionary for transcription
After creating your dictionary, you can apply it whenever you generate a new transcription.
- Click Create captions and choose Auto-transcribe.
- If Amazon Transcribe or Speechmatics is selected as your transcription provider, choose the custom dictionary you'd like to use.
- Make sure your preferred transcription provider is enabled under Team settings > Transcription.
- Complete the remaining transcription settings and click Submit.
Language variants
Custom dictionaries automatically work across regional variants of the same language. As long as the base ISO language matches, you only need to maintain a single dictionary.
For example, a dictionary created for English (UK) can also be used with English (US) or English (Global).
This behaviour applies whether you're:
- Auto transcribing from the Original captions tab
- Using an Automation template
Managing your custom dictionaries
The Custom dictionaries page lets you manage all dictionaries available to your team.
- Name: The name of the dictionary.
- Terms: The total number of terms it contains.
- Used in projects: The number of projects where the dictionary has been used.
- Language: The language assigned to the dictionary.
- Last updated: The date the dictionary was last modified.
The More actions (⋮) menu provides the following options:
- Edit: Update the dictionary contents. You can search, edit, select and bulk delete terms.
- Mark as default: Automatically apply this dictionary to all original caption transcriptions created from the Original captions tab or through Automation templates.
- Export phrases: Export all dictionary entries as a CSV file.
- Duplicate: Create a copy of the dictionary, useful when creating a variation.
- Delete: Permanently remove the dictionary.
