We believe Voice AI should be composable.

The orchestration layer shouldn't dictate your speech models. Your speech provider shouldn't dictate where you deploy. And your infrastructure shouldn't force trade-offs between performance, compliance, and flexibility.

That's why we're excited to announce Smallest AI's integration with Dograh. Developers can now use Smallest AI's speech models directly within Dograh's open-source voice agent platform, creating a fully customizable voice stack built around openness, choice, and performance.

With this integration, Dograh users can now power their agents using Smallest AI's speech stack:

• Lightning v3.1 for ultra-low latency text-to-speech

• Lightning v3.1 Pro for premium-quality voices

• Pulse for real-time speech-to-text

The integration allows developers to combine Dograh's visual workflow builder and open architecture with Smallest AI's speech models to create responsive, production-ready voice experiences.

What is Dograh?

Dograh is an open-source Voice AI platform designed for building and deploying voice agents.

Think of it as a visual workflow builder for voice applications - similar to how automation tools like n8n simplify workflow creation, but purpose-built for voice agents and AI calling.

Built by exited founders and YC alumni, Dograh is focused on making voice infrastructure more accessible, flexible, and transparent.

Key capabilities include:

• Visual workflow builder for voice agents

• Self-hosted and cloud deployment options

• Bring-your-own-model architecture

• Support for speech-to-speech workflows

• MCP-native integrations

• Open-source flexibility and compliance controls

For teams building customer-facing voice systems, Dograh offers an alternative approach to closed voice agent platforms, giving developers greater control over infrastructure, data handling, and model selection.

Why This Integration Matters

Voice agents are only as good as the quality and responsiveness of their speech stack.

A typical voice workflow looks something like this:

User Speech → Speech-to-Text → LLM → Text-to-Speech → User

Every component contributes to the overall experience, but speech models often have the largest impact on latency and conversational quality.

By integrating Smallest AI into Dograh, developers can now combine an open voice orchestration layer with fast, production-ready speech models.

Open Architecture without Vendor Lock-In

One of Dograh's core principles is flexibility.

Developers can:

• Use Smallest AI for both TTS and STT

• Use Smallest AI for only one part of the stack

• Mix and match providers depending on use case

This allows teams to optimize for quality, latency, cost, compliance, or deployment preferences.

Faster Development Cycles

Building voice agents traditionally requires stitching together multiple services, APIs, and infrastructure components. Dograh simplifies this process through a visual workflow builder, while Smallest AI provides plug-and-play speech capabilities.

The result is a faster path from prototype to production.

Available Smallest AI Models

Lightning v3.1

Lightning v3.1 is Smallest AI's low-latency text-to-speech model designed for real-time conversational applications.

It provides access to:

• 217 voices

• Optimized streaming performance

• Natural conversational speech

Lightning v3.1 Pro

Lightning v3.1 Pro is designed for teams that need higher-fidelity voice experiences.

It features:

• Premium voice pool

• Curated American, British, and Indian voices

• 44.1 kHz audio output

Pulse

Pulse is Smallest AI's real-time speech-to-text model.

It enables:

• Streaming transcription

• Low-latency speech recognition

• Real-time conversational workflows

Pulse serves as the foundation for voice agents that need immediate understanding of user speech before generating responses.

Getting Started

Setting up Smallest AI in Dograh takes only a few minutes.

Step 1: Open the Models Configuration Page

Navigate to the Models section within Dograh's workflow builder.

Step 2: Configure Smallest AI for Text-to-Speech

Under the Voice configuration:

• Select Smallest AI as the provider

• Choose either Lightning v3.1 or Lightning v3.1 Pro

• Select a voice

• Choose a language

• Adjust speech speed if required

Step 3: Configure Smallest AI for Speech-to-Text

Under the Transcriber configuration:

• Select Smallest AI as the provider

• Pulse will be selected automatically

• Choose your language

• Add your Smallest AI API key

Once configured, your Dograh agents can immediately begin using Smallest AI speech models.

For complete setup instructions, refer to the integration documentation: Docs Link

What Can You Build?

The combination of Dograh and Smallest AI enables a wide range of voice applications.

• AI Calling Agents

• Customer Support Automation

• Appointment Scheduling

• Sales and Lead Qualification

• Internal Enterprise Assistants

• Voice-Enabled MCP Agents

Building the Future of Open Voice AI

The voice AI ecosystem is evolving quickly, and we're seeing increasing demand for infrastructure that is both powerful and flexible. Open platforms allow developers to choose the components that work best for their use cases while maintaining control over deployment, data handling, and architecture.

By bringing Smallest AI's speech models to Dograh, we're making it easier for builders to create high-quality voice experiences without sacrificing flexibility. We're excited to see what the community builds with Dograh and Smallest AI together.

Get started today and power your next voice agent with Lightning TTS and Pulse STT.