Voice and answer quality

Two settings shape how your agent sounds and thinks: the Voice callers hear and the Answer quality. You find both under More settings → Add setting in the agent editor. If you add nothing, the agent uses Tegsoft's recommended settings, shown as "Now using: …" and kept up to date for you.

Choose a voice

1. Click Add setting and pick Voice.
2. Search, or filter by Female/Male, provider, This agent's languages or one language.
3. Type a sentence in "Type a sentence to hear it in a voice" (otherwise your First Message is used) and click Play on a few voices.
4. Click Use on your favourite (it then shows In use), then OK.

Each voice shows its cost in credits per minute. To go back to the recommended voice, click Remove and use the default on the setting's card.

Answer quality: how clever should the answers be?

Each level shows its credits per minute. With a level chosen, Tegsoft keeps the AI model behind it up to date, so your agent improves without you doing anything.

For technical users: choose a specific model

Choose a specific model instead pins an exact provider and model. Tegsoft then stops updating it for you. You can also set:

• Temperature: from Precise to Creative, 0.7 by default (not shown for every model).
• Max Response Tokens: 1024 by default, the ceiling on every channel. On calls, the shorter of this and the reply limit in Call Behaviour applies.

If a provider retires a pinned model, Tegsoft switches the agent to the matching answer quality so it keeps working.

Voice engine (TTS) and Speech to Text (STT)

• Voice engine (TTS) fine-tunes the engine behind your voice: model, speed and tone. The options depend on the voice, and they do not carry over if you choose a voice from another engine.
• Speech to Text (STT) chooses how the agent hears callers. With some answer qualities the model hears callers itself; the page then tells you the setting is used only if you change the model.