338 lines
10 KiB
JSON
338 lines
10 KiB
JSON
{
|
|
"family": "seed_vc",
|
|
"schema_version": 1,
|
|
"display_name": "Seed-VC",
|
|
"description": "Zero-shot voice conversion and singing voice conversion model for transferring timbre and style from reference audio, with low-latency realtime conversion and optional lightweight fine-tuning.",
|
|
"category": "voice_conversion",
|
|
"status": "supported",
|
|
"tasks": [
|
|
"vc",
|
|
"svc"
|
|
],
|
|
"modes": [
|
|
"offline"
|
|
],
|
|
"languages": [
|
|
"language_agnostic"
|
|
],
|
|
"capabilities": {
|
|
"vc": [
|
|
"speaker_reference"
|
|
],
|
|
"svc": [
|
|
"speaker_reference",
|
|
"singing"
|
|
]
|
|
},
|
|
"dependencies": [],
|
|
"runtime": {
|
|
"tags": [
|
|
"gguf"
|
|
]
|
|
},
|
|
"ui": {
|
|
"recommended_package": "seed_vc_mlx_q8_0",
|
|
"tags": [
|
|
"VC",
|
|
"GGUF"
|
|
],
|
|
"docs": [
|
|
"docs/models/seed_vc.md",
|
|
"docs/audio_tools.md",
|
|
"docs/gguf.md"
|
|
]
|
|
},
|
|
"options": {
|
|
"request": [
|
|
{
|
|
"name": "route",
|
|
"type": "enum",
|
|
"description": "Select the Seed-VC conversion route. Defaults to v2_vc for VC and v1_svc for SVC.",
|
|
"values": [
|
|
"v2_vc",
|
|
"v1_svc",
|
|
"v1_whisper_bigvgan_vc",
|
|
"v1_xlsr_hift_vc"
|
|
],
|
|
"required": false
|
|
},
|
|
{
|
|
"name": "length_adjust",
|
|
"type": "float",
|
|
"description": "Output duration multiplier; must be positive, default 1.0.",
|
|
"required": false,
|
|
"min": 0.0,
|
|
"default": 1.0
|
|
},
|
|
{
|
|
"name": "num_inference_steps",
|
|
"type": "int",
|
|
"description": "Diffusion steps; default 30.",
|
|
"required": false,
|
|
"min": 1,
|
|
"default": 30
|
|
},
|
|
{
|
|
"name": "inference_guidance_scale",
|
|
"type": "float",
|
|
"description": "V1 classifier-free guidance scale; default 0.7.",
|
|
"required": false,
|
|
"min": 0.0,
|
|
"default": 0.7
|
|
},
|
|
{
|
|
"name": "intelligibility_guidance_scale",
|
|
"type": "float",
|
|
"description": "V2 classifier-free guidance scale for source-content intelligibility; default 0.7.",
|
|
"required": false,
|
|
"min": 0.0,
|
|
"default": 0.7
|
|
},
|
|
{
|
|
"name": "similarity_guidance_scale",
|
|
"type": "float",
|
|
"description": "V2 classifier-free guidance scale for target-speaker similarity; default 0.7.",
|
|
"required": false,
|
|
"min": 0.0,
|
|
"default": 0.7
|
|
},
|
|
{
|
|
"name": "voice_anonymization",
|
|
"type": "bool",
|
|
"description": "Use randomized average-voice conditioning instead of target-speaker conditioning for V2 anonymization; default false.",
|
|
"required": false,
|
|
"default": false
|
|
},
|
|
{
|
|
"name": "seed",
|
|
"type": "int",
|
|
"description": "Seed for V1/V2 diffusion noise and HiFT stochastic source excitation; omitted requests choose a random seed.",
|
|
"required": false,
|
|
"min": 0
|
|
},
|
|
{
|
|
"name": "noise_path",
|
|
"type": "path",
|
|
"description": "Optional raw f32 noise file for deterministic V1/V2 diffusion noise and XLSR/HiFT source excitation.",
|
|
"required": false
|
|
},
|
|
{
|
|
"name": "f0_condition",
|
|
"type": "bool",
|
|
"description": "Enable V1 F0 conditioning for singing voice conversion; default false.",
|
|
"required": false,
|
|
"default": false
|
|
},
|
|
{
|
|
"name": "auto_f0_adjust",
|
|
"type": "bool",
|
|
"description": "Automatically adjust V1 source pitch toward the target pitch level; default false.",
|
|
"required": false,
|
|
"default": false
|
|
},
|
|
{
|
|
"name": "semitone_shift",
|
|
"type": "int",
|
|
"description": "V1 pitch shift in semitones for singing voice conversion; default 0.",
|
|
"required": false,
|
|
"default": 0
|
|
}
|
|
],
|
|
"session": [
|
|
{
|
|
"name": "weight_type",
|
|
"type": "enum",
|
|
"description": "Shared Seed-VC component weight storage type; default native, except RMVPE uses f32 unless overridden.",
|
|
"preset": "weight_type_full",
|
|
"required": false,
|
|
"default": "native"
|
|
}
|
|
],
|
|
"load": []
|
|
},
|
|
"package_defaults": {
|
|
"download": {
|
|
"kind": "huggingface_snapshot",
|
|
"repo": "audio-cpp/audio.cpp-gguf",
|
|
"revision": "main",
|
|
"gated": false
|
|
}
|
|
},
|
|
"packages": [
|
|
{
|
|
"id": "seed_vc_mlx_q8_0",
|
|
"display_name": "SeedVC-MLX Q8_0 GGUF",
|
|
"default": true,
|
|
"format": "gguf",
|
|
"precision": "q8_0",
|
|
"target_directory": "SeedVC-MLX-GGUF",
|
|
"files": [
|
|
"SeedVC-MLX-GGUF/seed-vc-mlx-q8_0.gguf"
|
|
],
|
|
"strip_prefix": "SeedVC-MLX-GGUF"
|
|
},
|
|
{
|
|
"id": "seed_vc_mlx_f16",
|
|
"display_name": "SeedVC-MLX F16 GGUF",
|
|
"format": "gguf",
|
|
"precision": "f16",
|
|
"target_directory": "SeedVC-MLX-GGUF",
|
|
"files": [
|
|
"SeedVC-MLX-GGUF/seed-vc-mlx-f16.gguf"
|
|
],
|
|
"strip_prefix": "SeedVC-MLX-GGUF"
|
|
},
|
|
{
|
|
"id": "seed_vc_mlx_orig",
|
|
"display_name": "SeedVC-MLX Original-Dtype GGUF",
|
|
"format": "gguf",
|
|
"precision": "orig",
|
|
"target_directory": "SeedVC-MLX-GGUF",
|
|
"files": [
|
|
"SeedVC-MLX-GGUF/seed-vc-mlx-orig.gguf"
|
|
],
|
|
"strip_prefix": "SeedVC-MLX-GGUF"
|
|
},
|
|
{
|
|
"id": "seed_vc_mlx_safetensors",
|
|
"display_name": "SeedVC-MLX Safetensors",
|
|
"format": "safetensors",
|
|
"precision": "native",
|
|
"target_directory": "SeedVC-MLX",
|
|
"files": [
|
|
"seed_vc_manifest.json",
|
|
"v2/ar.safetensors",
|
|
"v2/cfm.safetensors"
|
|
],
|
|
"download": {
|
|
"kind": "huggingface_snapshot",
|
|
"repo": "mlx-community/SeedVC-MLX"
|
|
}
|
|
}
|
|
],
|
|
"sources": [
|
|
{
|
|
"format": "gguf",
|
|
"roots": {
|
|
"model": ".",
|
|
"weights": "$gguf"
|
|
},
|
|
"files": {
|
|
"manifest": "model:seed_vc_manifest.json",
|
|
"v2_wrapper_config": "model:v2/vc_wrapper.json",
|
|
"astral_bsq32_config": "model:astral/bsq32.json",
|
|
"astral_bsq2048_config": "model:astral/bsq2048.json",
|
|
"v1_svc_config": "model:v1/svc.json",
|
|
"v1_whisper_bigvgan_config": "model:v1/whisper_bigvgan.json",
|
|
"v1_xlsr_hift_config": "model:v1/xlsr_hift.json",
|
|
"hift_config": "model:hift/config.json",
|
|
"bigvgan_22k_config": "model:bigvgan/v2_22khz_80band_256x/config.json",
|
|
"bigvgan_44k_config": "model:bigvgan/v2_44khz_128band_512x/config.json",
|
|
"whisper_small_config": "model:whisper-small/config.json",
|
|
"hubert_large_config": "model:hubert-large-ll60k/config.json",
|
|
"wav2vec2_xlsr_config": "model:wav2vec2-xls-r-300m/config.json"
|
|
},
|
|
"tensors": {
|
|
"v2_ar_weights": {
|
|
"source": "weights:",
|
|
"prefix": "v2_ar_weights"
|
|
},
|
|
"v2_cfm_weights": {
|
|
"source": "weights:",
|
|
"prefix": "v2_cfm_weights"
|
|
},
|
|
"v1_svc_weights": {
|
|
"source": "weights:",
|
|
"prefix": "v1_svc_weights"
|
|
},
|
|
"v1_whisper_bigvgan_weights": {
|
|
"source": "weights:",
|
|
"prefix": "v1_whisper_bigvgan_weights"
|
|
},
|
|
"v1_xlsr_hift_weights": {
|
|
"source": "weights:",
|
|
"prefix": "v1_xlsr_hift_weights"
|
|
},
|
|
"astral_bsq32_weights": {
|
|
"source": "weights:",
|
|
"prefix": "astral_bsq32_weights"
|
|
},
|
|
"astral_bsq2048_weights": {
|
|
"source": "weights:",
|
|
"prefix": "astral_bsq2048_weights"
|
|
},
|
|
"campplus_weights": {
|
|
"source": "weights:",
|
|
"prefix": "campplus_weights"
|
|
},
|
|
"rmvpe_weights": {
|
|
"source": "weights:",
|
|
"prefix": "rmvpe_weights"
|
|
},
|
|
"hift_weights": {
|
|
"source": "weights:",
|
|
"prefix": "hift_weights"
|
|
},
|
|
"bigvgan_22k_weights": {
|
|
"source": "weights:",
|
|
"prefix": "bigvgan_22k_weights"
|
|
},
|
|
"bigvgan_44k_weights": {
|
|
"source": "weights:",
|
|
"prefix": "bigvgan_44k_weights"
|
|
},
|
|
"whisper_small_weights": {
|
|
"source": "weights:",
|
|
"prefix": "whisper_small_weights"
|
|
},
|
|
"hubert_large_weights": {
|
|
"source": "weights:",
|
|
"prefix": "hubert_large_weights"
|
|
},
|
|
"wav2vec2_xlsr_weights": {
|
|
"source": "weights:",
|
|
"prefix": "wav2vec2_xlsr_weights"
|
|
}
|
|
}
|
|
},
|
|
{
|
|
"format": "safetensors",
|
|
"roots": {
|
|
"model": "."
|
|
},
|
|
"files": {
|
|
"manifest": "model:seed_vc_manifest.json",
|
|
"v2_wrapper_config": "model:v2/vc_wrapper.json",
|
|
"astral_bsq32_config": "model:astral/bsq32.json",
|
|
"astral_bsq2048_config": "model:astral/bsq2048.json",
|
|
"v1_svc_config": "model:v1/svc.json",
|
|
"v1_whisper_bigvgan_config": "model:v1/whisper_bigvgan.json",
|
|
"v1_xlsr_hift_config": "model:v1/xlsr_hift.json",
|
|
"hift_config": "model:hift/config.json",
|
|
"bigvgan_22k_config": "model:bigvgan/v2_22khz_80band_256x/config.json",
|
|
"bigvgan_44k_config": "model:bigvgan/v2_44khz_128band_512x/config.json",
|
|
"whisper_small_config": "model:whisper-small/config.json",
|
|
"hubert_large_config": "model:hubert-large-ll60k/config.json",
|
|
"wav2vec2_xlsr_config": "model:wav2vec2-xls-r-300m/config.json"
|
|
},
|
|
"tensors": {
|
|
"v2_ar_weights": "model:v2/ar.safetensors",
|
|
"v2_cfm_weights": "model:v2/cfm.safetensors",
|
|
"v1_svc_weights": "model:v1/svc.safetensors",
|
|
"v1_whisper_bigvgan_weights": "model:v1/whisper_bigvgan.safetensors",
|
|
"v1_xlsr_hift_weights": "model:v1/xlsr_hift.safetensors",
|
|
"astral_bsq32_weights": "model:astral/bsq32.safetensors",
|
|
"astral_bsq2048_weights": "model:astral/bsq2048.safetensors",
|
|
"campplus_weights": "model:campplus/model.safetensors",
|
|
"rmvpe_weights": "model:rmvpe/model.safetensors",
|
|
"hift_weights": "model:hift/model.safetensors",
|
|
"bigvgan_22k_weights": "model:bigvgan/v2_22khz_80band_256x/model.safetensors",
|
|
"bigvgan_44k_weights": "model:bigvgan/v2_44khz_128band_512x/model.safetensors",
|
|
"whisper_small_weights": "model:whisper-small/model.safetensors",
|
|
"hubert_large_weights": "model:hubert-large-ll60k/model.safetensors",
|
|
"wav2vec2_xlsr_weights": "model:wav2vec2-xls-r-300m/model.safetensors"
|
|
}
|
|
}
|
|
]
|
|
}
|