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
336 lines
9 KiB
JSON
336 lines
9 KiB
JSON
{
|
|
"family": "vibevoice_asr_streaming",
|
|
"display_name": "VibeVoice ASR Streaming",
|
|
"description": "Microsoft VibeVoice ASR Streaming models (7B and 1.5B) for chunked streaming speech-to-text with persistent decoder state.",
|
|
"category": "asr",
|
|
"status": "supported",
|
|
"tasks": [
|
|
"asr"
|
|
],
|
|
"modes": [
|
|
"offline",
|
|
"streaming"
|
|
],
|
|
"languages": [
|
|
"en",
|
|
"zh",
|
|
"es",
|
|
"pt",
|
|
"de",
|
|
"ja",
|
|
"ko",
|
|
"fr",
|
|
"ru",
|
|
"it"
|
|
],
|
|
"capabilities": {
|
|
"asr": [
|
|
"speaker_turns"
|
|
]
|
|
},
|
|
"options": {
|
|
"request": [
|
|
{
|
|
"name": "language",
|
|
"type": "string",
|
|
"description": "ASR language label; empty unless provided in the request.",
|
|
"required": false,
|
|
"default": ""
|
|
},
|
|
{
|
|
"name": "context",
|
|
"type": "string",
|
|
"description": "Extra context or hotwords injected into the streaming prompt.",
|
|
"required": false,
|
|
"default": ""
|
|
},
|
|
{
|
|
"name": "max_tokens",
|
|
"type": "int",
|
|
"description": "Maximum generated transcript tokens per chunk.",
|
|
"required": false,
|
|
"min": 1,
|
|
"default": 256
|
|
},
|
|
{
|
|
"name": "temperature",
|
|
"type": "float",
|
|
"description": "Sampling temperature; 0 uses deterministic decoding.",
|
|
"required": false,
|
|
"min": 0,
|
|
"default": 0
|
|
},
|
|
{
|
|
"name": "top_p",
|
|
"type": "float",
|
|
"description": "Nucleus sampling probability.",
|
|
"required": false,
|
|
"min": 0.001,
|
|
"max": 1,
|
|
"default": 1
|
|
},
|
|
{
|
|
"name": "top_k",
|
|
"type": "int",
|
|
"description": "Top-k sampling limit; default 50. Zero disables top-k filtering.",
|
|
"required": false,
|
|
"min": 0,
|
|
"default": 50
|
|
},
|
|
{
|
|
"name": "num_beams",
|
|
"type": "int",
|
|
"description": "Compatibility-only; this checkpoint ignores beam search and uses its incremental decoder.",
|
|
"required": false,
|
|
"min": 1,
|
|
"default": 1
|
|
},
|
|
{
|
|
"name": "repetition_penalty",
|
|
"type": "float",
|
|
"description": "Generation repetition penalty.",
|
|
"required": false,
|
|
"min": 0.001,
|
|
"default": 1
|
|
},
|
|
{
|
|
"name": "seed",
|
|
"type": "int",
|
|
"description": "Acoustic latent sampling seed; default 1234.",
|
|
"required": false,
|
|
"default": 1234
|
|
},
|
|
{
|
|
"name": "audio_chunk_mode",
|
|
"type": "enum",
|
|
"description": "Compatibility-only; ignored in offline and streaming modes. Processor chunk/lookahead settings determine segmentation.",
|
|
"values": [
|
|
"auto",
|
|
"fixed",
|
|
"vad",
|
|
"none"
|
|
],
|
|
"required": false,
|
|
"default": "auto"
|
|
},
|
|
{
|
|
"name": "audio_chunk_duration_sec",
|
|
"type": "float",
|
|
"description": "Compatibility-only; ignored in offline and streaming modes. Processor settings determine segmentation.",
|
|
"required": false,
|
|
"min": 0.001,
|
|
"default": 1200
|
|
}
|
|
],
|
|
"session": [
|
|
{
|
|
"name": "weight_type",
|
|
"type": "enum",
|
|
"description": "Shared component weight storage.",
|
|
"required": false,
|
|
"preset": "weight_type_full",
|
|
"default": "native"
|
|
},
|
|
{
|
|
"name": "tokenizer_weight_type",
|
|
"type": "enum",
|
|
"description": "Speech tokenizer weight storage; inherits weight_type when omitted.",
|
|
"required": false,
|
|
"preset": "weight_type_full"
|
|
},
|
|
{
|
|
"name": "connector_weight_type",
|
|
"type": "enum",
|
|
"description": "Speech connector weight storage; inherits weight_type when omitted.",
|
|
"required": false,
|
|
"preset": "weight_type_full"
|
|
},
|
|
{
|
|
"name": "decoder_weight_type",
|
|
"type": "enum",
|
|
"description": "Text decoder weight storage; inherits weight_type when omitted.",
|
|
"required": false,
|
|
"preset": "weight_type_full"
|
|
},
|
|
{
|
|
"name": "tokenizer_weight_context_mb",
|
|
"type": "int",
|
|
"description": "Speech tokenizer weight metadata context. Size in MiB; must be positive.",
|
|
"required": false,
|
|
"min": 1,
|
|
"default": 512
|
|
},
|
|
{
|
|
"name": "connector_weight_context_mb",
|
|
"type": "int",
|
|
"description": "Speech connector weight metadata context. Size in MiB; must be positive.",
|
|
"required": false,
|
|
"min": 1,
|
|
"default": 128
|
|
},
|
|
{
|
|
"name": "decoder_weight_context_mb",
|
|
"type": "int",
|
|
"description": "Text decoder weight metadata context. Default 4096 on 64-bit builds and 1024 on 32-bit builds. Size in MiB; must be positive.",
|
|
"required": false,
|
|
"min": 1
|
|
},
|
|
{
|
|
"name": "max_history_steps",
|
|
"type": "int",
|
|
"description": "Decoder history limit; zero uses the model position capacity. Must be nonnegative and not exceed that capacity.",
|
|
"required": false,
|
|
"min": 0,
|
|
"default": 0
|
|
}
|
|
],
|
|
"load": []
|
|
},
|
|
"runtime": {
|
|
"tags": [
|
|
"gguf"
|
|
]
|
|
},
|
|
"ui": {
|
|
"recommended_package": "vibevoice_asr_streaming_7b_q8_0",
|
|
"tags": [
|
|
"ASR",
|
|
"Streaming",
|
|
"GGUF"
|
|
],
|
|
"docs": [
|
|
"docs/asr.md",
|
|
"docs/gguf.md"
|
|
]
|
|
},
|
|
"package_defaults": {
|
|
"download": {
|
|
"kind": "huggingface_snapshot",
|
|
"repo": "audio-cpp/VibeVoice-ASR-Streaming-7B-GGUF",
|
|
"revision": "main",
|
|
"gated": false
|
|
}
|
|
},
|
|
"packages": [
|
|
{
|
|
"id": "vibevoice_asr_streaming_7b_q8_0",
|
|
"display_name": "VibeVoice ASR Streaming 7B Q8_0 GGUF",
|
|
"default": true,
|
|
"format": "gguf",
|
|
"precision": "q8_0",
|
|
"target_directory": "VibeVoice-ASR-Streaming-7B-GGUF",
|
|
"files": [
|
|
"vibevoice-asr-streaming-7b-q8_0.gguf"
|
|
]
|
|
},
|
|
{
|
|
"id": "vibevoice_asr_streaming_7b_bf16",
|
|
"display_name": "VibeVoice ASR Streaming 7B BF16 GGUF",
|
|
"format": "gguf",
|
|
"precision": "bf16",
|
|
"target_directory": "VibeVoice-ASR-Streaming-7B-GGUF",
|
|
"files": [
|
|
"vibevoice-asr-streaming-7b-bf16.gguf"
|
|
]
|
|
},
|
|
{
|
|
"id": "vibevoice_asr_streaming_7b_q4_k",
|
|
"display_name": "VibeVoice ASR Streaming 7B Q4_K GGUF",
|
|
"format": "gguf",
|
|
"precision": "q4_k",
|
|
"target_directory": "VibeVoice-ASR-Streaming-7B-GGUF",
|
|
"files": [
|
|
"vibevoice-asr-streaming-7b-q4_k.gguf"
|
|
]
|
|
},
|
|
{
|
|
"id": "vibevoice_asr_streaming_1_5b_q8_0",
|
|
"display_name": "VibeVoice ASR Streaming 1.5B Q8_0 GGUF",
|
|
"format": "gguf",
|
|
"precision": "q8_0",
|
|
"target_directory": "VibeVoice-ASR-Streaming-1.5B-GGUF",
|
|
"files": [
|
|
"vibevoice-asr-streaming-1.5b-q8_0.gguf"
|
|
],
|
|
"download": {
|
|
"kind": "huggingface_snapshot",
|
|
"repo": "christopherthompson81/VibeVoice-ASR-Streaming-1.5B-GGUF",
|
|
"revision": "main",
|
|
"gated": false
|
|
}
|
|
},
|
|
{
|
|
"id": "vibevoice_asr_streaming_1_5b_bf16",
|
|
"display_name": "VibeVoice ASR Streaming 1.5B BF16 GGUF",
|
|
"format": "gguf",
|
|
"precision": "bf16",
|
|
"target_directory": "VibeVoice-ASR-Streaming-1.5B-GGUF",
|
|
"files": [
|
|
"vibevoice-asr-streaming-1.5b-bf16.gguf"
|
|
],
|
|
"download": {
|
|
"kind": "huggingface_snapshot",
|
|
"repo": "christopherthompson81/VibeVoice-ASR-Streaming-1.5B-GGUF",
|
|
"revision": "main",
|
|
"gated": false
|
|
}
|
|
},
|
|
{
|
|
"id": "vibevoice_asr_streaming_1_5b_q4_k",
|
|
"display_name": "VibeVoice ASR Streaming 1.5B Q4_K GGUF",
|
|
"format": "gguf",
|
|
"precision": "q4_k",
|
|
"target_directory": "VibeVoice-ASR-Streaming-1.5B-GGUF",
|
|
"files": [
|
|
"vibevoice-asr-streaming-1.5b-q4_k.gguf"
|
|
],
|
|
"download": {
|
|
"kind": "huggingface_snapshot",
|
|
"repo": "christopherthompson81/VibeVoice-ASR-Streaming-1.5B-GGUF",
|
|
"revision": "main",
|
|
"gated": false
|
|
}
|
|
}
|
|
],
|
|
"sources": [
|
|
{
|
|
"format": "gguf",
|
|
"roots": {
|
|
"model": ".",
|
|
"weights": "$gguf"
|
|
},
|
|
"files": {
|
|
"config": "model:config.json",
|
|
"tokenizer_config": "model:tokenizer_config.json",
|
|
"tokenizer_json": "model:tokenizer.json",
|
|
"tokenizer_vocab": "model:vocab.json",
|
|
"tokenizer_merges": "model:merges.txt"
|
|
},
|
|
"optional_files": {
|
|
"preprocessor_config": "model:preprocessor_config.json"
|
|
},
|
|
"tensors": {
|
|
"model_weights": "weights:"
|
|
}
|
|
},
|
|
{
|
|
"format": "safetensors",
|
|
"roots": {
|
|
"model": "."
|
|
},
|
|
"files": {
|
|
"config": "model:config.json",
|
|
"tokenizer_config": "model:tokenizer_config.json",
|
|
"tokenizer_json": "model:tokenizer.json",
|
|
"tokenizer_vocab": "model:vocab.json",
|
|
"tokenizer_merges": "model:merges.txt"
|
|
},
|
|
"optional_files": {
|
|
"preprocessor_config": "model:preprocessor_config.json"
|
|
},
|
|
"tensors": {
|
|
"model_weights": "model:model.safetensors.index.json"
|
|
}
|
|
}
|
|
]
|
|
}
|