Integration: Microsoft Semantic Kernel (.NET)
<!-- metadata
type: queue
slug: 470-semantic-kernel-voice-plugin-dotnet
language: csharp
products: stt|tts
priority: user-request
-->
What this should show
A Deepgram plugin for Microsoft Semantic Kernel that exposes STT (pre-recorded transcription) and TTS (Aura) as [KernelFunction] attributes. Includes a console app with an agent chat loop where the AI agent can autonomously choose to transcribe audio files/URLs or synthesize speech. Demonstrates how .NET AI agent developers can integrate Deepgram voice capabilities into their Semantic Kernel pipelines.
Credentials likely needed
DEEPGRAM_API_KEY
OPENAI_API_KEY (or Azure OpenAI key for the Semantic Kernel LLM backend)
Original request:
What to build
A working example demonstrating Deepgram as a Microsoft Semantic Kernel plugin for .NET AI agents — enabling voice transcription as a kernel function that AI agents can invoke to process audio inputs.
Why this matters
Microsoft Semantic Kernel (53k+ stars) is the dominant .NET AI agent framework, used by enterprise teams building AI assistants and copilots. Developers building voice-enabled .NET agents need a reference integration showing Deepgram as a Semantic Kernel [KernelFunction]. There is currently no example of Deepgram + Semantic Kernel working together, leaving .NET AI agent developers without a clear path to add speech capabilities.
Suggested scope
- Language: C# / .NET 8
- Framework: Microsoft.SemanticKernel (latest stable)
- Deepgram APIs: Pre-recorded STT, TTS (Aura)
- What it does: Semantic Kernel plugin class with
[KernelFunction] attributes for TranscribeAudio (file/URL → text) and SpeakText (text → audio). Includes a sample agent chat loop that can process voice commands.
- Complexity: Medium — plugin class + sample console app
Acceptance criteria
Raised by the DX intelligence system.
Integration: Microsoft Semantic Kernel (.NET)
<!-- metadata
type: queue
slug: 470-semantic-kernel-voice-plugin-dotnet
language: csharp
products: stt|tts
priority: user-request
-->
What this should show
A Deepgram plugin for Microsoft Semantic Kernel that exposes STT (pre-recorded transcription) and TTS (Aura) as
[KernelFunction]attributes. Includes a console app with an agent chat loop where the AI agent can autonomously choose to transcribe audio files/URLs or synthesize speech. Demonstrates how .NET AI agent developers can integrate Deepgram voice capabilities into their Semantic Kernel pipelines.Credentials likely needed
DEEPGRAM_API_KEYOPENAI_API_KEY(or Azure OpenAI key for the Semantic Kernel LLM backend)Original request:
What to build
A working example demonstrating Deepgram as a Microsoft Semantic Kernel plugin for .NET AI agents — enabling voice transcription as a kernel function that AI agents can invoke to process audio inputs.
Why this matters
Microsoft Semantic Kernel (53k+ stars) is the dominant .NET AI agent framework, used by enterprise teams building AI assistants and copilots. Developers building voice-enabled .NET agents need a reference integration showing Deepgram as a Semantic Kernel
[KernelFunction]. There is currently no example of Deepgram + Semantic Kernel working together, leaving .NET AI agent developers without a clear path to add speech capabilities.Suggested scope
[KernelFunction]attributes forTranscribeAudio(file/URL → text) andSpeakText(text → audio). Includes a sample agent chat loop that can process voice commands.Acceptance criteria
Raised by the DX intelligence system.