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
342 lines
8.6 KiB
JSON
342 lines
8.6 KiB
JSON
{
|
|
"schema_version": 1,
|
|
"family": "fireredtts3",
|
|
"display_name": "FireRedTTS3",
|
|
"description": "FireRedTTS3 family for Base zero-shot voice cloning and Instruct instruction-conditioned TTS, voice design, semantic edit, and acoustic edit with Qwen3 autoregressive hidden-state generation, FireRed DiT latent patch flow, and RedAE waveform decoding.",
|
|
"category": "tts",
|
|
"status": "experimental",
|
|
"tasks": [
|
|
"tts",
|
|
"clone",
|
|
"design"
|
|
],
|
|
"modes": [
|
|
"offline"
|
|
],
|
|
"languages": [
|
|
"Chinese",
|
|
"English",
|
|
"Cantonese",
|
|
"Japanese",
|
|
"Korean",
|
|
"Spanish",
|
|
"French",
|
|
"Russian",
|
|
"Arabic",
|
|
"Turkish",
|
|
"Indonesian",
|
|
"Portuguese",
|
|
"Italian",
|
|
"Dutch",
|
|
"Vietnamese",
|
|
"German",
|
|
"Ukrainian",
|
|
"Thai",
|
|
"Polish",
|
|
"Romanian",
|
|
"Greek",
|
|
"Czech",
|
|
"Finnish",
|
|
"Hindi",
|
|
"ZH_Anhui",
|
|
"ZH_Fujian",
|
|
"ZH_Gansu",
|
|
"ZH_Guizhou",
|
|
"ZH_Hebei",
|
|
"ZH_Henan",
|
|
"ZH_Hubei",
|
|
"ZH_Hunan",
|
|
"ZH_Jiangxi",
|
|
"ZH_Liaoning",
|
|
"ZH_Minnan",
|
|
"ZH_Ningxia",
|
|
"ZH_Shaanxi",
|
|
"ZH_Shandong",
|
|
"ZH_Shanghai",
|
|
"ZH_Shanxi",
|
|
"ZH_Sichuan",
|
|
"ZH_Tianjin",
|
|
"ZH_Wenzhou",
|
|
"ZH_Wu",
|
|
"ZH_Yunnan"
|
|
],
|
|
"runtime": {
|
|
"tags": [
|
|
"gguf"
|
|
]
|
|
},
|
|
"dependencies": [],
|
|
"capabilities": {
|
|
"tts": [
|
|
"speaker_reference",
|
|
"style_control"
|
|
],
|
|
"clone": [
|
|
"speaker_reference",
|
|
"style_control"
|
|
],
|
|
"design": [
|
|
"voice_design"
|
|
]
|
|
},
|
|
"options": {
|
|
"request": [
|
|
{
|
|
"name": "template_name",
|
|
"type": "enum",
|
|
"description": "Instruct template; defaults to acoustic_edit with input audio, otherwise instruct_tts.",
|
|
"required": false,
|
|
"values": [
|
|
"instruct_tts",
|
|
"voice_design",
|
|
"semantic_edit",
|
|
"acoustic_edit"
|
|
]
|
|
},
|
|
{
|
|
"name": "reference_text",
|
|
"type": "string",
|
|
"description": "Transcript for the prompt audio.",
|
|
"required": false
|
|
},
|
|
{
|
|
"name": "instruct",
|
|
"type": "string",
|
|
"description": "FireRedTTS3 instruction for voice design or edit tasks.",
|
|
"required": false
|
|
},
|
|
{
|
|
"name": "language",
|
|
"type": "string",
|
|
"description": "Language tag used by FireRedTTS3.",
|
|
"required": false,
|
|
"default": "Chinese"
|
|
},
|
|
{
|
|
"name": "text_chunk_size",
|
|
"type": "int",
|
|
"description": "Maximum Unicode codepoints per long-form text chunk; default 80 when hanzi, kana and Hangul outnumber other letters and digits (except edits), otherwise 600.",
|
|
"required": false,
|
|
"min": 1
|
|
},
|
|
{
|
|
"name": "text_chunk_mode",
|
|
"type": "enum",
|
|
"description": "Framework text chunking mode.",
|
|
"values": [
|
|
"default",
|
|
"tag_aware",
|
|
"japanese",
|
|
"endline"
|
|
],
|
|
"required": false,
|
|
"default": "default"
|
|
},
|
|
{
|
|
"name": "num_inference_steps",
|
|
"type": "int",
|
|
"description": "FireRed DiT flow steps per generated latent patch.",
|
|
"required": false,
|
|
"min": 1,
|
|
"default": 10
|
|
},
|
|
{
|
|
"name": "guidance_scale",
|
|
"type": "float",
|
|
"description": "Positive guidance scale; defaults to 1.2 for voice-design/semantic-edit/acoustic-edit templates and 2 otherwise.",
|
|
"required": false,
|
|
"min": 0
|
|
},
|
|
{
|
|
"name": "stop_threshold",
|
|
"type": "float",
|
|
"description": "Stop probability threshold for AR patch generation.",
|
|
"required": false,
|
|
"min": 0.0,
|
|
"max": 1.0,
|
|
"default": 0.5
|
|
},
|
|
{
|
|
"name": "seed",
|
|
"type": "int",
|
|
"description": "Unsigned 32-bit sampling seed in [0, 4294967295]; defaults to 2 for voice design and 1234 otherwise.",
|
|
"required": false,
|
|
"min": 0,
|
|
"max": 4294967295
|
|
}
|
|
],
|
|
"session": [
|
|
{
|
|
"name": "weight_type",
|
|
"type": "enum",
|
|
"description": "FireRedTTS3 weight storage type; default native.",
|
|
"preset": "weight_type_full",
|
|
"required": false,
|
|
"default": "native"
|
|
},
|
|
{
|
|
"name": "graph_arena_mb",
|
|
"type": "int",
|
|
"description": "Main runtime graph arena size in MiB; default 1024.",
|
|
"required": false,
|
|
"min": 1,
|
|
"default": 1024
|
|
},
|
|
{
|
|
"name": "helper_graph_arena_mb",
|
|
"type": "int",
|
|
"description": "Helper graph arena size in MiB; default 256.",
|
|
"required": false,
|
|
"min": 1,
|
|
"default": 256
|
|
},
|
|
{
|
|
"name": "weight_context_mb",
|
|
"type": "int",
|
|
"description": "Weight loading context size in MiB; default 512.",
|
|
"required": false,
|
|
"min": 1,
|
|
"default": 512
|
|
},
|
|
{
|
|
"name": "reference_cache_slots",
|
|
"type": "int",
|
|
"description": "Prepared reference-audio cache slots; set 0 to disable reuse. Applies only to Base; Instruct does not use this cache.",
|
|
"required": false,
|
|
"min": 0,
|
|
"default": 4
|
|
},
|
|
{
|
|
"name": "mem_saver",
|
|
"type": "bool",
|
|
"description": "Release cached runtime graphs after each request to reduce peak VRAM; default false.",
|
|
"required": false,
|
|
"default": false
|
|
}
|
|
],
|
|
"load": []
|
|
},
|
|
"packages": [
|
|
{
|
|
"id": "fireredtts3_instruct_orig",
|
|
"display_name": "FireRedTTS3 Instruct Orig GGUF",
|
|
"format": "gguf",
|
|
"precision": "orig",
|
|
"target_directory": "FireRedTTS3-Instruct-GGUF",
|
|
"files": [
|
|
"FireRedTTS3-Instruct-GGUF/fireredtts3-instruct-orig.gguf"
|
|
],
|
|
"strip_prefix": "FireRedTTS3-Instruct-GGUF",
|
|
"download": {
|
|
"kind": "huggingface_snapshot",
|
|
"repo": "audio-cpp/audio.cpp-gguf",
|
|
"revision": "main",
|
|
"gated": false
|
|
}
|
|
},
|
|
{
|
|
"id": "fireredtts3_instruct_q8_0",
|
|
"display_name": "FireRedTTS3 Instruct Q8_0 GGUF",
|
|
"default": true,
|
|
"format": "gguf",
|
|
"precision": "q8_0",
|
|
"target_directory": "FireRedTTS3-Instruct-GGUF",
|
|
"files": [
|
|
"FireRedTTS3-Instruct-GGUF/fireredtts3-instruct-q8_0.gguf"
|
|
],
|
|
"strip_prefix": "FireRedTTS3-Instruct-GGUF",
|
|
"download": {
|
|
"kind": "huggingface_snapshot",
|
|
"repo": "audio-cpp/audio.cpp-gguf",
|
|
"revision": "main",
|
|
"gated": false
|
|
}
|
|
},
|
|
{
|
|
"id": "fireredtts3_base_orig",
|
|
"display_name": "FireRedTTS3 Base Orig GGUF",
|
|
"format": "gguf",
|
|
"precision": "orig",
|
|
"target_directory": "FireRedTTS3-Base-GGUF",
|
|
"files": [
|
|
"FireRedTTS3-Base-GGUF/fireredtts3-base-orig.gguf"
|
|
],
|
|
"strip_prefix": "FireRedTTS3-Base-GGUF",
|
|
"download": {
|
|
"kind": "huggingface_snapshot",
|
|
"repo": "audio-cpp/audio.cpp-gguf",
|
|
"revision": "main",
|
|
"gated": false
|
|
}
|
|
},
|
|
{
|
|
"id": "fireredtts3_base_q8_0",
|
|
"display_name": "FireRedTTS3 Base Q8_0 GGUF",
|
|
"format": "gguf",
|
|
"precision": "q8_0",
|
|
"target_directory": "FireRedTTS3-Base-GGUF",
|
|
"files": [
|
|
"FireRedTTS3-Base-GGUF/fireredtts3-base-q8_0.gguf"
|
|
],
|
|
"strip_prefix": "FireRedTTS3-Base-GGUF",
|
|
"download": {
|
|
"kind": "huggingface_snapshot",
|
|
"repo": "audio-cpp/audio.cpp-gguf",
|
|
"revision": "main",
|
|
"gated": false
|
|
}
|
|
}
|
|
],
|
|
"ui": {
|
|
"recommended_package": "fireredtts3_instruct_q8_0",
|
|
"tags": [
|
|
"TTS",
|
|
"Clone",
|
|
"Design",
|
|
"GGUF"
|
|
],
|
|
"docs": [
|
|
"docs/models/fireredtts3.md",
|
|
"docs/gguf.md"
|
|
]
|
|
},
|
|
"sources": [
|
|
{
|
|
"format": "gguf",
|
|
"roots": {
|
|
"model": ".",
|
|
"weights": "$gguf"
|
|
},
|
|
"files": {
|
|
"redae_config": "model:redae/config.json",
|
|
"tokenizer_json": "model:text_tokenizer/tokenizer.json",
|
|
"tokenizer_config": "model:text_tokenizer/tokenizer_config.json",
|
|
"vocab_json": "model:text_tokenizer/vocab.json"
|
|
},
|
|
"optional_files": {
|
|
"base_config": "model:fireredtts3_base/config.json",
|
|
"instruct_config": "model:fireredtts3_instruct/config.json"
|
|
},
|
|
"tensors": {
|
|
"redae_weights": {
|
|
"source": "weights:",
|
|
"prefix": "redae"
|
|
}
|
|
},
|
|
"optional_tensors": {
|
|
"base_weights": {
|
|
"source": "weights:",
|
|
"prefix": "base"
|
|
},
|
|
"campplus_weights": {
|
|
"source": "weights:",
|
|
"prefix": "campplus"
|
|
},
|
|
"instruct_weights": {
|
|
"source": "weights:",
|
|
"prefix": "instruct"
|
|
}
|
|
}
|
|
}
|
|
]
|
|
}
|