Personalities

Browse and preview AI voices for your agents, and manage custom pronunciation dictionaries. The Personalities section lets you audition voices, fine-tune how words are spoken, and ensure your agents sound exactly the way you want.

Browsing Voices

The Personalities page displays a card-based gallery of available voices from the voice library. Each voice card shows the voice name, a sample text excerpt, and a set of descriptive badges covering language, gender, age, accent, emotions, and category. This makes it easy to scan through the collection and identify voices that match the persona you have in mind for your agent.

The gallery updates as new voices become available. Each card is designed to give you a quick, at-a-glance summary so you can narrow down your choices before previewing audio.

Filtering

To help you find the right voice quickly, the gallery includes a comprehensive set of filters. You can filter by:

• Language: Narrow results to voices that support a specific language.
• Gender: Choose from male, female, or neutral voices.
• Age: Select young, adult, or elderly voice profiles.
• Emotion: Find voices with specific emotional tones such as calm, friendly, or authoritative.
• Accent: Filter by regional accent to match your target audience.
• Category: Browse by voice category for themed or specialised use cases.

All filter options are dynamically populated based on the available voice catalogue, so you will only see values that match at least one voice in the current collection.

Previewing

Click the play button on any voice card to hear a short audio preview. This lets you evaluate the voice's tone, pace, and overall character before committing to it. Previews use a standard sample phrase so you can compare voices consistently.

Once you have found a voice you like, click the copy button to copy the voice ID to your clipboard. You can then paste this ID into your agent's voice configuration to assign the voice.

Custom Pronunciation

Navigate to Personalities > Pronunciation to manage custom pronunciation dictionaries. These dictionaries let you override how specific words or phrases are spoken, which is invaluable for brand names, technical terms, acronyms, or any word the default text-to-speech engine mispronounces.

Creating a Dictionary

To create a pronunciation dictionary, click Create Dictionary and provide a name for the collection. Each dictionary supports phonetic notation in either IPA (International Phonetic Alphabet) or CMU (Carnegie Mellon University phonetic notation) format. Choose the format you are most comfortable with; both produce high-quality results.

You can create multiple dictionaries to organise pronunciations by topic, client, or use case. For example, you might have one dictionary for medical terminology and another for your company's product names.

Adding Words

Once a dictionary exists, click Add Word to add individual entries. For each word, provide the original spelling and the phonetic representation in your chosen notation format. The phonetic spelling tells the text-to-speech engine exactly how to articulate the word, overriding its default pronunciation.

You can add as many words as needed to a single dictionary. Entries can be edited or removed at any time without affecting other words in the collection.

Testing Pronunciation

Before deploying a pronunciation dictionary to a live agent, use the built-in test pronunciation feature. Select any available voice, enter custom text that includes the words you have defined, and click Test to hear the result. This lets you verify that each word sounds correct in context and with the specific voice your agent will use.

If a word does not sound right, adjust the phonetic spelling in the dictionary and test again. Iterate until you are satisfied with the output; it is much easier to fine-tune pronunciation here than to debug it during a live call.