michaeljosiah 

Frame-based, real-time voice AI pipeline framework for .NET. Inspired by Pipecat. Provides core abstractions: Frame, FrameProcessor, Pipeline, PipelineRunner.

Vendor-neutral STT and TTS abstractions for the Voxa pipeline. Defines ISpeechToTextEngine, ITextToSpeechEngine, and the generic SpeechToTextProcessor / TextToSpeechProcessor that any vendor engine plugs into.

Adapter that wraps any Microsoft.Agents.AI AIAgent as a Voxa FrameProcessor. Bridges transcribed user input into the agent and pumps streamed text + tool calls back into the pipeline.

WebSocket source and sink processors for the Voxa pipeline. Operates over any System.Net.WebSockets.WebSocket — works with ASP.NET Core, ClientWebSocket, and any host that surfaces a WebSocket.

OpenAI Whisper STT and OpenAI TTS engines for the Voxa pipeline. Implements ISpeechToTextEngine + ITextToSpeechEngine via the REST API — pair with the generic processors from Voxa.Speech.Abstractions.

Composite STT+LLM+TTS+VAD processor for the Voxa pipeline framework, backed by the Azure Voice Live API. The same processor speaks Azure OpenAI Realtime and OpenAI Realtime by changing the endpoint.

Azure Speech Services STT and TTS engines for the Voxa pipeline. Implements ISpeechToTextEngine and ITextToSpeechEngine — pair with the generic SpeechToTextProcessor / TextToSpeechProcessor from... More information

ElevenLabs text-to-speech engine for the Voxa pipeline. Implements ITextToSpeechEngine via the streaming REST API — pair with the generic TextToSpeechProcessor from Voxa.Speech.Abstractions.

Mistral text-to-speech engine for the Voxa pipeline. Implements ITextToSpeechEngine via Mistral's Voxtral-TTS audio API — pair with the generic TextToSpeechProcessor from Voxa.Speech.Abstractions.

ML-based voice activity detection for the Voxa pipeline using the Silero VAD ONNX model. Drop-in replacement for SilenceGateProcessor when energy-based filtering isn't enough — handles noisy environments and... More information

Fluent ASP.NET Core integration for Voxa. Provides MapVoxaVoice + a VoicePipelineBuilder so MAF-using apps can drop in a voice pipeline with a single fluent expression: app.MapVoxaVoice("/voice", v =>... More information

Composite STT+LLM+TTS+VAD processor for the Voxa pipeline framework, backed by the OpenAI Realtime API. Full-duplex streaming voice with server-side turn detection — the closest C# equivalent to Pipecat's... More information

Test helpers for Voxa pipelines: WAV file processors, capturing processors, in-memory transports.

OpenTelemetry tracing helpers for Voxa pipelines. Provides the TracingProcessor (drop in anywhere to emit per-frame Activity spans) on top of the public Voxa ActivitySource.

Local/offline text-to-speech for the Voxa pipeline using Piper. No API key, no network after the first-run voice/binary download — develop without cloud accounts, run zero-cost CI conversations, deploy... More information

Local/offline speech-to-text for the Voxa pipeline using whisper.cpp (via Whisper.net). No API key, no network after the first-run model download — develop without cloud accounts, run zero-cost CI... More information

Local/offline high-quality text-to-speech for the Voxa pipeline using Kokoro-82M (Apache-2.0) on ONNX Runtime, in-process. No API key, no network after the first-run model download. The quality tier of Voxa's... More information

Audio-stage abstractions for the Voxa pipeline — the seams that sit on the mic path before the VAD. Defines IEchoCanceller (VRT-003, acoustic echo cancellation for barge-in over speakers) and IAudioEnhancer... More information

A Model Context Protocol (MCP) server that gives any MCP-aware agent a voice and ears via Voxa's keyless local speech tier: voxa_speak (text → WAV) and voxa_transcribe (WAV → text). Runs over stdio; install... More information

Batteries-included Voxa package. One package reference: AddVoxa(cfg) + MapVoxaVoice("/voice").UseDefaults() gives a working voice bot with STT, TTS, VAD, and an OpenAI chat agent — no knowledge of frames... More information