Voxa.Audio.Diarization.Onnx 0.7.2-alpha

.NET 10.0 This package targets .NET 10.0. The package is compatible with this framework or higher.
This is a prerelease version of Voxa.Audio.Diarization.Onnx. 
dotnet add package Voxa.Audio.Diarization.Onnx --version 0.7.2-alpha
                    


                    

                
NuGet\Install-Package Voxa.Audio.Diarization.Onnx -Version 0.7.2-alpha
This command is intended to be used within the Package Manager Console in Visual Studio, as it uses the NuGet module's version of Install-Package. 
<PackageReference Include="Voxa.Audio.Diarization.Onnx" Version="0.7.2-alpha" />
For projects that support PackageReference, copy this XML node into the project file to reference the package. 
<PackageVersion Include="Voxa.Audio.Diarization.Onnx" Version="0.7.2-alpha" />
                    


                            Directory.Packages.props
                        

                    

                
<PackageReference Include="Voxa.Audio.Diarization.Onnx" />
                    


                            Project file
For projects that support Central Package Management (CPM), copy this XML node into the solution Directory.Packages.props file to version the package. 
paket add Voxa.Audio.Diarization.Onnx --version 0.7.2-alpha
                    


                    

                
#r "nuget: Voxa.Audio.Diarization.Onnx, 0.7.2-alpha"
#r directive can be used in F# Interactive and Polyglot Notebooks. Copy this into the interactive tool or source code of the script to reference the package. 
#:package Voxa.Audio.Diarization.Onnx@0.7.2-alpha
#:package directive can be used in C# file-based apps starting in .NET 10 preview 4. Copy this into a .cs file before any lines of code to reference the package. 
#addin nuget:?package=Voxa.Audio.Diarization.Onnx&version=0.7.2-alpha&prerelease
                    


                            Install as a Cake Addin
                        

                    

                
#tool nuget:?package=Voxa.Audio.Diarization.Onnx&version=0.7.2-alpha&prerelease
                    


                            Install as a Cake Tool

Voxa.Audio.Diarization.Onnx

The reference ONNX implementations for Voxa speaker diarization (VLS-005 WS2) — the model-backed engines that plug into the seams shipped by Voxa.Audio.Diarization and run on the shared Voxa.Audio.Onnx host.

PyannoteOnnxSegmentation (ISpeakerSegmentation)

Speaker segmentation / speech-region detection backed by pyannote segmentation-3.0 (MIT, © CNRS).

The reason it's a clean ONNX-on-host fit: the model's SincNet front-end is inside the graph, so it takes raw 16 kHz audio — there's no external STFT/mel to implement or get numerically wrong. This engine only:

  1. frames the audio into the model's sliding windows (geometry read from the ONNX metadata — nothing hard-coded, so a re-export carries its own parameters),
  2. runs the graph on the OnnxModelHost (weights load once, process-wide), and
  3. decodes the powerset output into absolute-time speech regions in pure C# (PowersetSegmentationDecoder — a faithful port of the sherpa-onnx reference: powerset→speech per frame, Hamming-weighted overlap-add of the windows, onset/offset binarisation). That decoding is unit-tested on synthetic logits — no model needed.
var host = new OnnxModelHost();
var path = await cache.ResolveAsync(PyannoteSegmentationCatalog.Model, ct); // pinned, SHA-256-verified
var segmentation = new PyannoteOnnxSegmentation(path, host);
var pipeline = new DiarizationPipeline(segmentation, embedding); // embedding = WeSpeaker (follow-up)

Pinned model

PyannoteSegmentationCatalog.Model pins the sherpa-onnx export of pyannote-3.0 — an ungated, plain .onnx (the official pyannote/segmentation-3.0 repo is HF-gated; this byte-identical mirror isn't), so it resolves through VoxaModelCache with no archive step or HF token. SHA-256 sourced from the real artifact.

Status / follow-ups

  • Shipped: the segmentation engine + the pinned model + the pure decoder (unit-tested) + a LocalModels smoke test that runs the real model end-to-end.
  • Follow-ups: the ISpeakerEmbedding ONNX impl (WeSpeaker, CC-BY-4.0 — needs an Fbank front-end) to complete the full diarization stack, and the voxa transcribe --diarize CLI consumer.
Product Compatible and additional computed target framework versions.
.NET net10.0 is compatible.  net10.0-android was computed.  net10.0-browser was computed.  net10.0-ios was computed.  net10.0-maccatalyst was computed.  net10.0-macos was computed.  net10.0-tvos was computed.  net10.0-windows was computed. 
Compatible target framework(s)
Included target framework(s) (in package)
Learn more about Target Frameworks and .NET Standard.

NuGet packages

This package is not used by any NuGet packages.

GitHub repositories

This package is not used by any popular GitHub repositories.

Version Downloads Last Updated
0.7.2-alpha 86 7/10/2026
0.7.1-alpha 64 7/10/2026
0.7.0-alpha 66 7/10/2026
0.6.0-alpha 81 6/22/2026