
High-quality TTS is a core part of a great voice agent, but the cost of experimentation is a major bottleneck. Finding the right voice for your use case often means cycling through different models and styles, which can get expensive before you’ve even shipped a prototype.
For one week, we’re removing that barrier.
We've partnered with MiniMax to give the Vapi community free, unlimited, and unrestricted access to their entire TTS library. This is a chance to test their production-grade voices on our sub-500ms infrastructure without any cost constraints.
Here’s what the Vapi x MiniMax API week includes:
VAPIMINIMAX10
To help you get the most out of the week, here are some recommendations for configuring their models on our platform.
The MiniMax Speech-02-turbo model supports over 40 languages. While the auto setting for language_boost works well, we recommend selecting a specific language tag for specialized use cases (e.g., Thai, Japanese) to achieve the best results.
Path: Assistant → Voice → Language
Path: Workflow → Global Voice → Language
Path: Conversation → Voice → Language
To ensure proper pronunciation of numbers, emails, and dates in customer service scenarios, enable text normalization. This feature is on by default, and we recommend keeping it active.
Path: Voice → Additional Configuration → Text Normalization
If you don’t need a custom clone, we have integrated a default library of high-quality, pre-trained MiniMax voices. Simply select your preferred option from the Voice List in your settings to get started quickly.
This is a great opportunity to experiment with new voices and build something new.
Get started on the Vapi Dashboard
We’re excited to see what you build.