Gabber

Gabber: Realtime AI Voice, Text, Memory, and Tool-Calling

Overview

Gabber serves as an advanced inference engine specifically designed for real-time AI applications and personas. It allows users to create multi-modal AI agents capable of speech, text, visual perception, context switching, tool calling, and memory retention. This platform empowers developers and businesses to build AI solutions that can intuitively interact with users and perform complex tasks.

Features

Create Multi-Modal AI Agents

Gabber enables the development of AI platforms utilizing open-source models, effectively facilitating the creation of AI agents, applications, and personas. Users can either engage in quick inference or construct full-fledged custom AI personas tailored to specific use cases.

Build Your AI Experience

Quick Inference vs. Full Persona

Users have the flexibility to choose between:

• Quick Inference: This option provides rapid access to pre-defined features and behaviors, ideal for immediate use cases.
• Build Full Persona: This approach allows developers to create entirely custom AI personas that meet distinct user needs.

Model Selection

Gabber offers users direct access to a variety of state-of-the-art models, enhancing the ease of integration through websockets, SDKs, and APIs. Some available models include:

• Cartesia Voice: A low-latency voice option that allows for clone-like real-time voice inference at a cost of $3 per hour.
• Orpheus Voice: An emotive, high-quality voice option perfect for applications requiring emotional expression, available for $1 per hour.
• NSFW LLM: An unrestricted language model that allows for content freedom, free to use.
• SFW LLM: A safe, moderated language model suitable for all audiences, also available for free.

Simplifying Integration

The era of backend complexity has concluded with Gabber's user-friendly approach to API calls. By wrapping API keys in user-linked tokens, developers can securely utilize SDKs directly on the frontend without compromising safety.

Generating User Tokens

The platform offers straightforward methods for generating user tokens, enabling specific usage limits such as voice usage and token allocation.

Example code for generating a user token:

const token = await generateToken({
  user_id: "[email protected]",
  limits: {
    voice_seconds: 100,
    tokens: 10000
  }
});

SDK and API Integration

Gabber provides a seamless way to start real-time sessions and integrate its capabilities into applications with ease, showcased by example code that emphasizes user integration:

const { api } = useApi();

api.realtime.startRealtimeSession({
  usageToken: token
});

<RealtimeSessionEngineProvider>
  <VoiceChat />
</RealtimeSessionEngineProvider>

Usage Tracking

As part of its comprehensive suite, Gabber includes features for tracking usage metrics, which is vital for understanding usage patterns and ensuring effective management:

const usage = await getUsage();
console.log(usage);
// {
//   voice_minutes: 45,
//   messages: 328
// }

Conclusion

Gabber stands out as a versatile and powerful tool for building real-time AI applications. Its ability to create agents that can speak, text, and switch contexts makes it ideal for a variety of industries, including customer support, education, and interactive entertainment. With both a rich feature set and user-friendly integration processes, Gabber is especially useful for developers looking to leverage AI technology efficiently and effectively in their projects.

Overall, Gabber equips users with the tools necessary to create engaging and intelligent AI applications, unlocking new potential for interaction and user experience. Whether one is working on a small-scale personal project or a large enterprise solution, Gabber offers the capability to build AI that feels truly alive.