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
277 lines
7.9 KiB
JSON
277 lines
7.9 KiB
JSON
{
|
|
"schema_version": 1,
|
|
"family": "neutts",
|
|
"display_name": "NeuTTS",
|
|
"description": "NeuTTS is an English text-to-speech family from Neuphonic. The current 2E package uses a Qwen3-style autoregressive speech-token backbone, NeuCodec waveform decoding, built-in speaker prompts, and emotion-token control.",
|
|
"category": "tts",
|
|
"status": "supported",
|
|
"tasks": [
|
|
"tts"
|
|
],
|
|
"modes": [
|
|
"offline",
|
|
"streaming"
|
|
],
|
|
"languages": [
|
|
"en"
|
|
],
|
|
"runtime": {
|
|
"tags": [
|
|
"gguf",
|
|
"stream"
|
|
]
|
|
},
|
|
"capabilities": {
|
|
"tts": [
|
|
"built_in_voices",
|
|
"emotion_control",
|
|
"long_form"
|
|
]
|
|
},
|
|
"options": {
|
|
"request": [
|
|
{
|
|
"name": "voice_id",
|
|
"type": "enum",
|
|
"description": "Built-in NeuTTS speaker prompt; default emily.",
|
|
"values": [
|
|
"dave",
|
|
"emily",
|
|
"greta",
|
|
"jo",
|
|
"juliette",
|
|
"mateo",
|
|
"paul",
|
|
"sophie",
|
|
"steven"
|
|
],
|
|
"required": false,
|
|
"default": "emily"
|
|
},
|
|
{
|
|
"name": "emotion",
|
|
"type": "enum",
|
|
"description": "Optional emotion token inserted between reference text and target text; neutral inserts no emotion token.",
|
|
"values": [
|
|
"angry",
|
|
"disgusted",
|
|
"sad",
|
|
"happy",
|
|
"fearful",
|
|
"neutral",
|
|
"surprised"
|
|
],
|
|
"required": false,
|
|
"default": "neutral"
|
|
},
|
|
{
|
|
"name": "max_tokens",
|
|
"type": "int",
|
|
"description": "Maximum generated speech-token count for the autoregressive generator; 0 uses the remaining model context, default 0. Total generation context is capped at min(checkpoint context, 2048).",
|
|
"required": false,
|
|
"min": 0,
|
|
"default": 0
|
|
},
|
|
{
|
|
"name": "min_tokens",
|
|
"type": "int",
|
|
"description": "Minimum generated speech-token count before EOS may stop generation; default 50.",
|
|
"required": false,
|
|
"min": 0,
|
|
"default": 50
|
|
},
|
|
{
|
|
"name": "temperature",
|
|
"type": "float",
|
|
"description": "Autoregressive sampling temperature; must be > 0, default 1.0.",
|
|
"required": false,
|
|
"min": 0.0,
|
|
"default": 1.0
|
|
},
|
|
{
|
|
"name": "top_k",
|
|
"type": "int",
|
|
"description": "Autoregressive top-k sampling limit; default 50.",
|
|
"required": false,
|
|
"min": 1,
|
|
"default": 50
|
|
},
|
|
{
|
|
"name": "seed",
|
|
"type": "int",
|
|
"description": "Autoregressive sampling seed; omitted requests choose a random seed.",
|
|
"required": false,
|
|
"min": 0
|
|
},
|
|
{
|
|
"name": "text_chunk_mode",
|
|
"type": "enum",
|
|
"description": "Framework text chunking mode for long-form synthesis.",
|
|
"values": [
|
|
"default",
|
|
"tag_aware",
|
|
"japanese",
|
|
"endline"
|
|
],
|
|
"required": false,
|
|
"default": "default"
|
|
},
|
|
{
|
|
"name": "text_chunk_size",
|
|
"type": "int",
|
|
"description": "Maximum Unicode codepoints per long-form text chunk; default 600.",
|
|
"required": false,
|
|
"min": 1,
|
|
"default": 600
|
|
}
|
|
],
|
|
"session": [
|
|
{
|
|
"name": "weight_type",
|
|
"type": "enum",
|
|
"description": "Shared matmul weight storage type for the backbone and codec decoder; default native.",
|
|
"preset": "weight_type_full",
|
|
"required": false,
|
|
"default": "native"
|
|
},
|
|
{
|
|
"name": "generator_weight_type",
|
|
"type": "enum",
|
|
"description": "Generator weight storage; inherits explicit weight_type, otherwise F32 on CPU and native on other backends.",
|
|
"preset": "weight_type_full",
|
|
"required": false
|
|
},
|
|
{
|
|
"name": "codec_weight_type",
|
|
"type": "enum",
|
|
"description": "NeuCodec decoder matmul weight storage type; defaults to weight_type when set, otherwise native.",
|
|
"preset": "weight_type_full",
|
|
"required": false
|
|
},
|
|
{
|
|
"name": "codec_conv_weight_type",
|
|
"type": "enum",
|
|
"description": "NeuCodec convolution weight storage type; default native.",
|
|
"preset": "weight_type_conv",
|
|
"required": false,
|
|
"default": "native"
|
|
},
|
|
{
|
|
"name": "runtime_graph_arena_mb",
|
|
"type": "int",
|
|
"description": "Reusable ggml graph arena size in MiB for NeuTTS runtime graphs; default 1024.",
|
|
"required": false,
|
|
"min": 1,
|
|
"default": 1024
|
|
}
|
|
],
|
|
"load": []
|
|
},
|
|
"package_defaults": {
|
|
"download": {
|
|
"kind": "huggingface_snapshot",
|
|
"repo": "audio-cpp/audio.cpp-gguf",
|
|
"revision": "main",
|
|
"gated": false
|
|
}
|
|
},
|
|
"packages": [
|
|
{
|
|
"id": "neutts_2e_orig",
|
|
"display_name": "NeuTTS 2E Original-Precision GGUF",
|
|
"default": true,
|
|
"format": "gguf",
|
|
"precision": "orig",
|
|
"target_directory": "NeuTTS-2E-GGUF",
|
|
"files": [
|
|
"NeuTTS-2E-GGUF/neutts-2e-orig.gguf"
|
|
],
|
|
"strip_prefix": "NeuTTS-2E-GGUF"
|
|
}
|
|
],
|
|
"dependencies": [],
|
|
"ui": {
|
|
"recommended_package": "neutts_2e_orig",
|
|
"tags": [
|
|
"TTS",
|
|
"GGUF",
|
|
"Stream"
|
|
],
|
|
"docs": [
|
|
"docs/tts.md",
|
|
"docs/models/neutts.md",
|
|
"docs/gguf.md"
|
|
]
|
|
},
|
|
"sources": [
|
|
{
|
|
"format": "gguf",
|
|
"roots": {
|
|
"model": ".",
|
|
"weights": "$gguf"
|
|
},
|
|
"files": {
|
|
"config": "model:config.json",
|
|
"generation_config": "model:generation_config.json",
|
|
"tokenizer_json": "model:tokenizer.json",
|
|
"tokenizer_config": "model:tokenizer_config.json",
|
|
"chat_template": "model:chat_template.jinja",
|
|
"codec_config": "model:neucodec_config.json",
|
|
"codec_preprocessor_config": "model:neucodec_preprocessor_config.json",
|
|
"speaker_text_dave": "model:samples/dave.txt",
|
|
"speaker_text_emily": "model:samples/emily.txt",
|
|
"speaker_text_greta": "model:samples/greta.txt",
|
|
"speaker_text_jo": "model:samples/jo.txt",
|
|
"speaker_text_juliette": "model:samples/juliette.txt",
|
|
"speaker_text_mateo": "model:samples/mateo.txt",
|
|
"speaker_text_paul": "model:samples/paul.txt",
|
|
"speaker_text_sophie": "model:samples/sophie.txt",
|
|
"speaker_text_steven": "model:samples/steven.txt"
|
|
},
|
|
"tensors": {
|
|
"backbone": {
|
|
"source": "weights:",
|
|
"prefix": "backbone"
|
|
},
|
|
"codec": {
|
|
"source": "weights:",
|
|
"prefix": "codec"
|
|
},
|
|
"speaker_prompts": {
|
|
"source": "weights:",
|
|
"prefix": "speaker_prompts"
|
|
}
|
|
}
|
|
},
|
|
{
|
|
"format": "safetensors",
|
|
"roots": {
|
|
"model": ".",
|
|
"codec": "../NeuCodec"
|
|
},
|
|
"files": {
|
|
"config": "model:config.json",
|
|
"generation_config": "model:generation_config.json",
|
|
"tokenizer_json": "model:tokenizer.json",
|
|
"tokenizer_config": "model:tokenizer_config.json",
|
|
"chat_template": "model:chat_template.jinja",
|
|
"codec_config": "codec:config.json",
|
|
"codec_preprocessor_config": "codec:preprocessor_config.json",
|
|
"speaker_text_dave": "model:samples/dave.txt",
|
|
"speaker_text_emily": "model:samples/emily.txt",
|
|
"speaker_text_greta": "model:samples/greta.txt",
|
|
"speaker_text_jo": "model:samples/jo.txt",
|
|
"speaker_text_juliette": "model:samples/juliette.txt",
|
|
"speaker_text_mateo": "model:samples/mateo.txt",
|
|
"speaker_text_paul": "model:samples/paul.txt",
|
|
"speaker_text_sophie": "model:samples/sophie.txt",
|
|
"speaker_text_steven": "model:samples/steven.txt"
|
|
},
|
|
"tensors": {
|
|
"backbone": "model:model.safetensors",
|
|
"codec": "codec:model.safetensors",
|
|
"speaker_prompts": "model:samples/speaker_prompts.safetensors"
|
|
}
|
|
}
|
|
]
|
|
}
|