voice_audio_multi/model_specs/plapre.json

{
  "schema_version": 1,
  "family": "plapre",
  "display_name": "Plapre 3",
  "description": "Danish speech synthesis with a Llama autoregressive decoder and Kanade audio codec. Streaming emits completed text chunks, not incremental codec frames.",
  "category": "tts",
  "status": "supported",
  "tasks": [
    "tts",
    "clone"
  ],
  "modes": [
    "offline",
    "streaming"
  ],
  "languages": [
    "da"
  ],
  "capabilities": {
    "tts": [
      "long_form",
      "speaker_reference",
      "built_in_voices"
    ],
    "clone": [
      "speaker_reference",
      "long_form"
    ]
  },
  "runtime": {
    "tags": [
      "gguf",
      "stream"
    ]
  },
  "options": {
    "request": [
      {
        "name": "voice_id",
        "type": "enum",
        "description": "Built-in speaker.",
        "required": false,
        "default": "tor",
        "values": [
          "tor",
          "ida",
          "liv",
          "ask",
          "kaj"
        ]
      },
      {
        "name": "reference_text",
        "type": "string",
        "description": "Transcript of the reference audio, used for voice cloning.",
        "required": false,
        "default": ""
      },
      {
        "name": "x_vector_only_mode",
        "type": "bool",
        "description": "Use only the reference speaker embedding, without the reference transcript/audio-token prefix.",
        "required": false,
        "default": false
      },
      {
        "name": "word_duration_frames",
        "type": "string",
        "description": "JSON array of per-word integer durations, from 0 to 399 frames at 25 Hz. Duration-controlled text is generated in one chunk.",
        "required": false
      },
      {
        "name": "context_text",
        "type": "string",
        "description": "Previous utterance for prosodic context.",
        "required": false
      },
      {
        "name": "context_audio_file",
        "type": "audio_path",
        "description": "Optional previous-utterance WAV accompanying context_text.",
        "required": false
      },
      {
        "name": "same_speaker",
        "type": "bool",
        "description": "Whether context_text belongs to the target speaker.",
        "required": false,
        "default": true
      },
      {
        "name": "pronunciations",
        "type": "string",
        "description": "Ordered JSON array of pronunciation references: [{\"word\":\"morgen\",\"audio_file\":\"morgen.wav\"}]. Up to 10 references, 60 codes each and 200 total codes are used.",
        "required": false
      },
      {
        "name": "edit_start_frame",
        "type": "int",
        "description": "Start of the masked span in voice_ref, in 25-Hz codec frames. Supply both edit boundaries and the complete desired text.",
        "required": false,
        "min": 0
      },
      {
        "name": "edit_end_frame",
        "type": "int",
        "description": "Exclusive end of the masked span in voice_ref. Editing runs in one text chunk.",
        "required": false,
        "min": 0
      },
      {
        "name": "max_tokens",
        "type": "int",
        "description": "Maximum generated tokens per text chunk; each audio token represents 40 ms.",
        "required": false,
        "default": 500,
        "min": 1
      },
      {
        "name": "min_tokens",
        "type": "int",
        "description": "Minimum generated tokens before a stop token can be selected.",
        "required": false,
        "default": 0,
        "min": 0
      },
      {
        "name": "temperature",
        "type": "float",
        "description": "Sampling temperature; zero selects greedy decoding.",
        "required": false,
        "default": 0.8,
        "min": 0
      },
      {
        "name": "top_p",
        "type": "float",
        "description": "Nucleus sampling probability in (0, 1].",
        "required": false,
        "default": 0.95,
        "min": 0,
        "max": 1
      },
      {
        "name": "top_k",
        "type": "int",
        "description": "Top-k sampling; -1 or zero disables filtering.",
        "required": false,
        "default": 50,
        "min": -1
      },
      {
        "name": "seed",
        "type": "int",
        "description": "Sampling and vocoder noise seed.",
        "required": false,
        "default": 0,
        "min": 0
      },
      {
        "name": "text_chunk_size",
        "type": "int",
        "description": "Maximum Unicode codepoints per long-form text chunk.",
        "required": false,
        "default": 300,
        "min": 1
      },
      {
        "name": "text_chunk_mode",
        "type": "enum",
        "description": "Framework text chunking policy.",
        "required": false,
        "default": "default",
        "values": [
          "default",
          "tag_aware",
          "japanese",
          "endline"
        ]
      },
      {
        "name": "silence_duration_sec",
        "type": "float",
        "description": "Silence inserted between text chunks.",
        "required": false,
        "default": 0.1,
        "min": 0
      }
    ],
    "session": [],
    "load": []
  },
  "package_defaults": {
    "download": {
      "kind": "huggingface_snapshot",
      "repo": "audio-cpp/audio.cpp-gguf",
      "revision": "main",
      "gated": false
    }
  },
  "packages": [
    {
      "id": "plapre_3_bf16",
      "display_name": "Plapre 3 BF16",
      "format": "gguf",
      "precision": "bf16",
      "default": true,
      "target_directory": "Plapre-3-GGUF",
      "files": [
        "Plapre-3-GGUF/plapre-3-bf16.gguf"
      ],
      "strip_prefix": "Plapre-3-GGUF"
    },
    {
      "id": "plapre_3_f32",
      "display_name": "Plapre 3 F32",
      "format": "gguf",
      "precision": "f32",
      "default": false,
      "target_directory": "Plapre-3-GGUF",
      "files": [
        "Plapre-3-GGUF/plapre-3-f32.gguf"
      ],
      "strip_prefix": "Plapre-3-GGUF"
    }
  ],
  "dependencies": [],
  "ui": {
    "recommended_package": "plapre_3_bf16",
    "builtin_voices": [
      "tor",
      "ida",
      "liv",
      "ask",
      "kaj"
    ],
    "default_voice": "tor",
    "tags": [
      "TTS",
      "GGUF"
    ],
    "docs": [
      "docs/models/plapre.md"
    ]
  },
  "sources": [
    {
      "format": "gguf",
      "roots": {
        "model": ".",
        "weights": "$gguf"
      },
      "files": {
        "config": "model:config.json",
        "kanade_config": "model:kanade_config.json",
        "speakers": "model:speakers.json",
        "tokenizer_json": "model:tokenizer.json",
        "tokenizer_config": "model:tokenizer_config.json"
      },
      "tensors": {
        "ar": "weights:",
        "speaker_projection": {
          "source": "weights:",
          "prefix": "speaker_proj"
        },
        "kanade": {
          "source": "weights:",
          "prefix": "kanade"
        },
        "wavlm": {
          "source": "weights:",
          "prefix": "wavlm"
        },
        "hift": {
          "source": "weights:",
          "prefix": "hift"
        }
      }
    }
  ]
}