Voxa.Audio.Onnx
Shared ONNX Runtime session host for the Voxa local-speech tier (VLS-006). Owns the InferenceSession lifecycle so every ONNX model (VAD, TTS, ASR, enhancement) loads its weights once per (path, device) on a process-wide cache, behind a CPU-default execution-provider seam — GPU is opt-in via a user-added Microsoft.ML.OnnxRuntime.Gpu/DirectML package and never bundled by Voxa. Ships the host + the device seam + the model-descriptor shape; consuming engines bring their own pinned catalogs.
Activity
- Latest release
- 2mo ago
- Total releases
- 4
- Cadence
- ~daily
- Last 12 months
- 4
Details
- License
- MIT
- First release
- Jun 22, 2026
| Version | Released | |
|---|---|---|
0.7.2-alpha
pre
|
0.7.2-alpha
pre
Dependencies (2)
|
|
0.7.1-alpha
pre
|
0.7.1-alpha
pre
Dependencies (2)
|
|
0.7.0-alpha
pre
|
0.7.0-alpha
pre
Dependencies (2)
|
|
0.6.0-alpha
pre
|
0.6.0-alpha
pre
Dependencies (2)
|