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
240 lines
8 KiB
JSON
240 lines
8 KiB
JSON
{
|
|
"schema_version": 1,
|
|
"family": "echo_tts",
|
|
"display_name": "Echo-TTS",
|
|
"description": "Echo-TTS is an English zero-shot voice-cloning TTS model packaged for audio.cpp. A 2.8B diffusion transformer generates 80-D latents in PCA space which the Fish S1-DAC decodes to 44.1 kHz audio. Generation is a fixed 29.72 s window (640 latents).",
|
|
"category": "tts",
|
|
"status": "experimental",
|
|
"tasks": [
|
|
"tts",
|
|
"clone"
|
|
],
|
|
"modes": [
|
|
"offline"
|
|
],
|
|
"languages": [
|
|
"en"
|
|
],
|
|
"runtime": {
|
|
"tags": [
|
|
"gguf"
|
|
]
|
|
},
|
|
"capabilities": {
|
|
"clone": [
|
|
"speaker_reference"
|
|
]
|
|
},
|
|
"options": {
|
|
"request": [
|
|
{
|
|
"name": "text_chunk_size",
|
|
"type": "int",
|
|
"description": "Maximum Unicode codepoints per text chunk; must be positive, default 300.",
|
|
"required": false,
|
|
"min": 1,
|
|
"default": 300
|
|
},
|
|
{
|
|
"name": "text_guidance_scale",
|
|
"type": "float",
|
|
"description": "Classifier-free guidance scale on the text condition (independent-cfg only). Advertised range is recommended, not enforced by the parser.",
|
|
"required": false,
|
|
"min": 0,
|
|
"default": 3
|
|
},
|
|
{
|
|
"name": "speaker_guidance_scale",
|
|
"type": "float",
|
|
"description": "Classifier-free guidance scale on the speaker condition (independent-cfg only). Advertised range is recommended, not enforced by the parser.",
|
|
"required": false,
|
|
"min": 0,
|
|
"default": 8
|
|
},
|
|
{
|
|
"name": "guidance_scale",
|
|
"type": "float",
|
|
"description": "Classifier-free guidance scale on the text and speaker condition (join-cfg only). Advertised range is recommended, not enforced by the parser.",
|
|
"required": false,
|
|
"min": 0,
|
|
"default": 8
|
|
},
|
|
{
|
|
"name": "t_schedule_mode",
|
|
"type": "enum",
|
|
"description": "RF Euler timestep schedule: linear or sway; default linear.",
|
|
"values": [
|
|
"linear",
|
|
"sway"
|
|
],
|
|
"required": false,
|
|
"default": "linear"
|
|
},
|
|
{
|
|
"name": "sway_coeff",
|
|
"type": "float",
|
|
"description": "Sway Sampling coefficient; must be finite and produce strictly decreasing timesteps. Default -1.0.",
|
|
"required": false,
|
|
"default": -1.0
|
|
},
|
|
{
|
|
"name": "num_inference_steps",
|
|
"type": "int",
|
|
"description": "Euler sampler steps. Default 40 for linear timestep schedule, 12 for sway.",
|
|
"required": false,
|
|
"min": 1,
|
|
"default": 40
|
|
},
|
|
{
|
|
"name": "truncation_factor",
|
|
"type": "float",
|
|
"description": "Initial-noise truncation factor. Advertised range is recommended, not enforced by the parser.",
|
|
"required": false,
|
|
"min": 0,
|
|
"max": 1,
|
|
"default": 0.8
|
|
},
|
|
{
|
|
"name": "speaker_kv_scale",
|
|
"type": "float",
|
|
"description": "Force-speaker KV scaling. 1.0 disables; 1.5 is the upstream default when enabled. Advertised range is recommended, not parser-enforced; zero is invalid for reciprocal scaling.",
|
|
"required": false,
|
|
"min": 1,
|
|
"default": 1
|
|
},
|
|
{
|
|
"name": "seed",
|
|
"type": "int",
|
|
"description": "RNG seed for the initial latent. Negative signed-integer values are treated as zero.",
|
|
"required": false,
|
|
"default": 0
|
|
},
|
|
{
|
|
"name": "reference_duration_sec",
|
|
"type": "float",
|
|
"required": false,
|
|
"description": "Trim the speaker reference to at most this many seconds before encoding. Shorter references are cheaper and often clone better; upstream's guidance favours around 10 s. Defaults to 15 s. Values above the trained maximum of 297.1 s are clamped. Set per request, or as a session default from CLI or server config. Must be positive.",
|
|
"default": 15
|
|
},
|
|
{
|
|
"name": "max_duration_sec",
|
|
"type": "float",
|
|
"description": "Cap the generation window, in seconds (up to 29.7215 s, the trained maximum; larger values are clamped to it). Quantised down to a whole latent frame of 46.44 ms. Left unset, the window is estimated per chunk from text length and widened automatically if the utterance does not finish inside it, which is substantially cheaper for short text. Must be positive; rounded to at least one latent frame.",
|
|
"required": false,
|
|
"max": 29.7215
|
|
},
|
|
{
|
|
"name": "guidance_interval",
|
|
"type": "int",
|
|
"description": "Refresh the two unconditional CFG lanes only every Nth step inside the guidance window, reusing the previous guidance correction in between. 1 reproduces upstream exactly. Worth raising to 2 at 30+ steps, where the correction is still measured 8-10 times across the guided phase (roughly 1.3x fewer denoiser evaluations); not recommended below ~20 steps, where it is already coarsely sampled and staleness shows up as degraded timbre and text adherence.",
|
|
"required": false,
|
|
"min": 1,
|
|
"default": 1
|
|
}
|
|
],
|
|
"session": [
|
|
{
|
|
"name": "reference_duration_sec",
|
|
"type": "float",
|
|
"required": false,
|
|
"description": "Trim the speaker reference to at most this many seconds before encoding. Shorter references are cheaper and often clone better; upstream's guidance favours around 10 s. Defaults to 15 s. Values above the trained maximum of 297.1 s are clamped. Set per request, or as a session default from CLI or server config. Must be positive.",
|
|
"default": 15
|
|
},
|
|
{
|
|
"name": "reference_cache_slots",
|
|
"type": "int",
|
|
"required": false,
|
|
"description": "How many encoded speaker references to keep. Encoding is linear in reference length, so reusing a voice across requests avoids repeating it. 0 disables caching; default 4."
|
|
},
|
|
{
|
|
"name": "mem_saver",
|
|
"type": "bool",
|
|
"required": false,
|
|
"description": "Release Fish S1-DAC decode graph space before encoding reference speaker audio and encode graph space after. Avoids having both allocations resident simulaneously at the cost of rebuilding the graphs when a speaker cache-miss occurs.",
|
|
"default": false
|
|
},
|
|
{
|
|
"name": "cfg_mode",
|
|
"type": "enum",
|
|
"description": "Type of classifier free guidance used.",
|
|
"values": [
|
|
"independent",
|
|
"joint"
|
|
],
|
|
"required": false,
|
|
"default": "independent"
|
|
}
|
|
],
|
|
"load": []
|
|
},
|
|
"packages": [
|
|
{
|
|
"id": "echo_tts_q8_0",
|
|
"display_name": "Echo-TTS Q8_0 GGUF",
|
|
"default": true,
|
|
"format": "gguf",
|
|
"precision": "q8_0",
|
|
"target_directory": "Echo-TTS-GGUF",
|
|
"files": [
|
|
"echo-tts-q8_0.gguf"
|
|
],
|
|
"download": {
|
|
"kind": "huggingface_snapshot",
|
|
"repo": "dignome/Echo-TTS",
|
|
"revision": "919681c9906c88963352d5f96e0f732ddd3989ce",
|
|
"gated": false
|
|
}
|
|
},
|
|
{
|
|
"id": "echo_tts_f16",
|
|
"display_name": "Echo-TTS F16 GGUF",
|
|
"format": "gguf",
|
|
"precision": "f16",
|
|
"target_directory": "Echo-TTS-GGUF",
|
|
"files": [
|
|
"echo-tts-f16.gguf"
|
|
],
|
|
"download": {
|
|
"kind": "huggingface_snapshot",
|
|
"repo": "dignome/Echo-TTS",
|
|
"revision": "919681c9906c88963352d5f96e0f732ddd3989ce",
|
|
"gated": false
|
|
}
|
|
}
|
|
],
|
|
"dependencies": [],
|
|
"ui": {
|
|
"recommended_package": "echo_tts_q8_0",
|
|
"tags": [
|
|
"TTS",
|
|
"Clone",
|
|
"GGUF"
|
|
],
|
|
"docs": [
|
|
"docs/models/echo_tts.md"
|
|
]
|
|
},
|
|
"sources": [
|
|
{
|
|
"format": "gguf",
|
|
"roots": {
|
|
"model": ".",
|
|
"weights": "$gguf"
|
|
},
|
|
"tensors": {
|
|
"dit_weights": {
|
|
"source": "weights:",
|
|
"prefix": "dit_weights"
|
|
},
|
|
"pca": {
|
|
"source": "weights:",
|
|
"prefix": "pca"
|
|
},
|
|
"codec_weights": {
|
|
"source": "weights:",
|
|
"prefix": "ae"
|
|
}
|
|
}
|
|
}
|
|
]
|
|
}
|