voice_audio_multi/model_specs/pulsevad.json
Fabio c36510c21e
Some checks failed
Linux Build / Linux cpu (push) Has been cancelled
Linux Build / Linux cpu-clang (push) Has been cancelled
Linux Build / Linux vulkan (push) Has been cancelled
mac-build / macOS CPU (push) Has been cancelled
Nix Build / Nix (cpu) (push) Has been cancelled
Nix Build / Nix (vulkan) (push) Has been cancelled
windows-build / Windows CPU (push) Has been cancelled
Initial import: audio.cpp with Orange Pi documentation
2026-11-04 19:05:22 +01:00

148 lines
3.2 KiB
JSON

{
"schema_version": 1,
"family": "pulsevad",
"display_name": "PulseVAD",
"description": "Speech activity detection for 16 kHz mono audio with 2.1K student and 81K teacher weights.",
"category": "audio_tools",
"status": "supported",
"tasks": [
"vad"
],
"modes": [
"offline"
],
"languages": [
"language_agnostic"
],
"capabilities": {
"vad": [
"speech_segments"
]
},
"runtime": {
"tags": [
"gguf"
]
},
"options": {
"request": [
{
"name": "threshold",
"type": "float",
"description": "Speech probability threshold.",
"required": false,
"default": 0.5,
"min": 0,
"max": 1
},
{
"name": "hop_size_samples",
"type": "int",
"description": "Window step in samples; 1600 samples is 100 ms at 16 kHz.",
"required": false,
"default": 1600,
"min": 1
},
{
"name": "min_speech_duration_ms",
"type": "int",
"description": "Minimum speech segment duration in milliseconds. Not applied to the short-input path (<3200 samples), which classifies the whole clip by threshold.",
"required": false,
"default": 100,
"min": 0
},
{
"name": "min_silence_duration_ms",
"type": "int",
"description": "Minimum silence before closing a speech segment, in milliseconds.",
"required": false,
"default": 100,
"min": 0
}
],
"session": [
{
"name": "weight_type",
"type": "enum",
"description": "Weight storage override; native preserves the packaged tensor types.",
"values": [
"native",
"f32",
"f16",
"bf16"
],
"required": false,
"default": "native"
}
],
"load": []
},
"package_defaults": {
"download": {
"kind": "huggingface_snapshot",
"repo": "audio-cpp/audio.cpp-gguf",
"revision": "main",
"gated": false
}
},
"packages": [
{
"id": "pulsevad_2_1k_f32",
"display_name": "PulseVAD 2.1K GGUF F32",
"default": true,
"format": "gguf",
"precision": "f32",
"target_directory": "PulseVAD-GGUF",
"files": [
"PulseVAD-GGUF/pulsevad-2.1k-f32.gguf"
],
"strip_prefix": "PulseVAD-GGUF"
},
{
"id": "pulsevad_81k_f32",
"display_name": "PulseVAD 81K GGUF F32",
"format": "gguf",
"precision": "f32",
"target_directory": "PulseVAD-GGUF",
"files": [
"PulseVAD-GGUF/pulsevad-81k-f32.gguf"
],
"strip_prefix": "PulseVAD-GGUF"
}
],
"dependencies": [],
"ui": {
"tags": [
"VAD",
"GGUF"
],
"docs": [
"docs/models/pulsevad.md"
],
"recommended_package": "pulsevad_2_1k_f32"
},
"sources": [
{
"format": "gguf",
"roots": {
"model": ".",
"weights": "$gguf"
},
"tensors": {
"weights": {
"source": "weights:",
"prefix": "weights"
}
}
},
{
"format": "safetensors",
"roots": {
"model": "."
},
"tensors": {
"weights": "model:pulsevad-f32.safetensors"
}
}
]
}