{
 "paper": {
  "title": "SemanticAudio: Audio Generation and Editing in Semantic Space",
  "short_title": "SemanticAudio",
  "authors": "Anonymous Authors",
  "code_url": "https://github.com/semanticaudio1/semanticaudio1.github.io/tree/main/code",
  "abstract_summary": "SemanticAudio decouples text-to-audio generation into semantic planning and acoustic synthesis. A Semantic Planner first predicts compact frame-level semantic latents that describe what sound events happen and when; an Acoustic Synthesizer then renders those plans into high-fidelity audio. The same semantic space enables training-free, text-guided FlowEdit for targeted audio editing.",
  "figures": {
   "semanticaudio": "assets/figures/semanticaudio_framework.png",
   "edit_framework": "assets/figures/semantic_flowedit_framework.png",
   "table1_audiocaps_generation": "assets/figures/table1_audiocaps_generation.png",
   "table2_ttabench": "assets/figures/table2_ttabench.png",
   "table4_editing": "assets/figures/table4_editing.png"
  }
 },
 "audiocaps": [
  {
   "id": "YLvMA1Wcgu3w",
   "index": 1,
   "caption": "A man speaking as frogs croak and crickets chirp while a motorboat engine runs alongside several plastic clacks and clanging",
   "source": "AudioCaps test split",
   "audio": {
    "semantic": {
     "label": "SemanticAudio",
     "path": "assets/audio/audiocaps/semantic/YLvMA1Wcgu3w.wav"
    },
    "tangoflux": {
     "label": "TangoFlux",
     "path": "assets/audio/audiocaps/tangoflux/YLvMA1Wcgu3w.wav"
    },
    "resonate": {
     "label": "Resonate",
     "path": "assets/audio/audiocaps/resonate/YLvMA1Wcgu3w.wav"
    },
    "base": {
     "label": "Base Model",
     "path": "assets/audio/audiocaps/base/YLvMA1Wcgu3w.wav"
    }
   }
  },
  {
   "id": "YnmLMLgWPmFM",
   "index": 2,
   "caption": "A motorboat engine running as water splashes and a man shouts in the background followed by birds chirping in the distance",
   "source": "AudioCaps test split",
   "audio": {
    "semantic": {
     "label": "SemanticAudio",
     "path": "assets/audio/audiocaps/semantic/YnmLMLgWPmFM.wav"
    },
    "tangoflux": {
     "label": "TangoFlux",
     "path": "assets/audio/audiocaps/tangoflux/YnmLMLgWPmFM.wav"
    },
    "resonate": {
     "label": "Resonate",
     "path": "assets/audio/audiocaps/resonate/YnmLMLgWPmFM.wav"
    },
    "base": {
     "label": "Base Model",
     "path": "assets/audio/audiocaps/base/YnmLMLgWPmFM.wav"
    }
   }
  },
  {
   "id": "YQRtuOWWya30",
   "index": 3,
   "caption": "Food and oil sizzling followed by metal rattling as a man is talking while guitar music plays and water trickles in the background",
   "source": "AudioCaps test split",
   "audio": {
    "semantic": {
     "label": "SemanticAudio",
     "path": "assets/audio/audiocaps/semantic/YQRtuOWWya30.wav"
    },
    "tangoflux": {
     "label": "TangoFlux",
     "path": "assets/audio/audiocaps/tangoflux/YQRtuOWWya30.wav"
    },
    "resonate": {
     "label": "Resonate",
     "path": "assets/audio/audiocaps/resonate/YQRtuOWWya30.wav"
    },
    "base": {
     "label": "Base Model",
     "path": "assets/audio/audiocaps/base/YQRtuOWWya30.wav"
    }
   }
  },
  {
   "id": "YCO6-i8NLbeo",
   "index": 4,
   "caption": "A man talking followed by a goat baaing then a metal gate sliding as ducks quack and wind blows into a microphone",
   "source": "AudioCaps test split",
   "audio": {
    "semantic": {
     "label": "SemanticAudio",
     "path": "assets/audio/audiocaps/semantic/YCO6-i8NLbeo.wav"
    },
    "tangoflux": {
     "label": "TangoFlux",
     "path": "assets/audio/audiocaps/tangoflux/YCO6-i8NLbeo.wav"
    },
    "resonate": {
     "label": "Resonate",
     "path": "assets/audio/audiocaps/resonate/YCO6-i8NLbeo.wav"
    },
    "base": {
     "label": "Base Model",
     "path": "assets/audio/audiocaps/base/YCO6-i8NLbeo.wav"
    }
   }
  },
  {
   "id": "YbygBWUkpaC8",
   "index": 5,
   "caption": "A man talking as another man speaks in the background followed by birds chirping and leaves rustling while wind blows into a microphone",
   "source": "AudioCaps test split",
   "audio": {
    "semantic": {
     "label": "SemanticAudio",
     "path": "assets/audio/audiocaps/semantic/YbygBWUkpaC8.wav"
    },
    "tangoflux": {
     "label": "TangoFlux",
     "path": "assets/audio/audiocaps/tangoflux/YbygBWUkpaC8.wav"
    },
    "resonate": {
     "label": "Resonate",
     "path": "assets/audio/audiocaps/resonate/YbygBWUkpaC8.wav"
    },
    "base": {
     "label": "Base Model",
     "path": "assets/audio/audiocaps/base/YbygBWUkpaC8.wav"
    }
   }
  },
  {
   "id": "YJsoBpL86R5U",
   "index": 6,
   "caption": "A man baaing followed by a goat baaing as a child and a woman talk while metal pots clank in the background",
   "source": "AudioCaps test split",
   "audio": {
    "semantic": {
     "label": "SemanticAudio",
     "path": "assets/audio/audiocaps/semantic/YJsoBpL86R5U.wav"
    },
    "tangoflux": {
     "label": "TangoFlux",
     "path": "assets/audio/audiocaps/tangoflux/YJsoBpL86R5U.wav"
    },
    "resonate": {
     "label": "Resonate",
     "path": "assets/audio/audiocaps/resonate/YJsoBpL86R5U.wav"
    },
    "base": {
     "label": "Base Model",
     "path": "assets/audio/audiocaps/base/YJsoBpL86R5U.wav"
    }
   }
  },
  {
   "id": "YAMQei29haCw",
   "index": 7,
   "caption": "A power tool drill running before turning off and on as a man is talking and wood is clanking and scrapping in the background",
   "source": "AudioCaps test split",
   "audio": {
    "semantic": {
     "label": "SemanticAudio",
     "path": "assets/audio/audiocaps/semantic/YAMQei29haCw.wav"
    },
    "tangoflux": {
     "label": "TangoFlux",
     "path": "assets/audio/audiocaps/tangoflux/YAMQei29haCw.wav"
    },
    "resonate": {
     "label": "Resonate",
     "path": "assets/audio/audiocaps/resonate/YAMQei29haCw.wav"
    },
    "base": {
     "label": "Base Model",
     "path": "assets/audio/audiocaps/base/YAMQei29haCw.wav"
    }
   }
  },
  {
   "id": "Y6i5eQOpFk_U",
   "index": 8,
   "caption": "Faucet water flowing as plastic clanks and a man speaks followed by another plastic clank and a kid laughing in the background",
   "source": "AudioCaps test split",
   "audio": {
    "semantic": {
     "label": "SemanticAudio",
     "path": "assets/audio/audiocaps/semantic/Y6i5eQOpFk_U.wav"
    },
    "tangoflux": {
     "label": "TangoFlux",
     "path": "assets/audio/audiocaps/tangoflux/Y6i5eQOpFk_U.wav"
    },
    "resonate": {
     "label": "Resonate",
     "path": "assets/audio/audiocaps/resonate/Y6i5eQOpFk_U.wav"
    },
    "base": {
     "label": "Base Model",
     "path": "assets/audio/audiocaps/base/Y6i5eQOpFk_U.wav"
    }
   }
  },
  {
   "id": "YTdl9SmBbRnA",
   "index": 9,
   "caption": "A large motor vehicle engine accelerates and then slows and idles, while an adult male speaks in the foreground",
   "source": "AudioCaps test split",
   "audio": {
    "semantic": {
     "label": "SemanticAudio",
     "path": "assets/audio/audiocaps/semantic/YTdl9SmBbRnA.wav"
    },
    "tangoflux": {
     "label": "TangoFlux",
     "path": "assets/audio/audiocaps/tangoflux/YTdl9SmBbRnA.wav"
    },
    "resonate": {
     "label": "Resonate",
     "path": "assets/audio/audiocaps/resonate/YTdl9SmBbRnA.wav"
    },
    "base": {
     "label": "Base Model",
     "path": "assets/audio/audiocaps/base/YTdl9SmBbRnA.wav"
    }
   }
  },
  {
   "id": "YoZaEHkfh5Eg",
   "index": 10,
   "caption": "A series of electronic dings while a vehicle horn honks twice followed by a plastic click while a dog barks in the distance",
   "source": "AudioCaps test split",
   "audio": {
    "semantic": {
     "label": "SemanticAudio",
     "path": "assets/audio/audiocaps/semantic/YoZaEHkfh5Eg.wav"
    },
    "tangoflux": {
     "label": "TangoFlux",
     "path": "assets/audio/audiocaps/tangoflux/YoZaEHkfh5Eg.wav"
    },
    "resonate": {
     "label": "Resonate",
     "path": "assets/audio/audiocaps/resonate/YoZaEHkfh5Eg.wav"
    },
    "base": {
     "label": "Base Model",
     "path": "assets/audio/audiocaps/base/YoZaEHkfh5Eg.wav"
    }
   }
  },
  {
   "id": "YXi6V0LGvqoo",
   "index": 11,
   "caption": "A dog whimpering followed by a dog growling and barking as metal jingles and footsteps squeak on hard surface",
   "source": "AudioCaps test split",
   "audio": {
    "semantic": {
     "label": "SemanticAudio",
     "path": "assets/audio/audiocaps/semantic/YXi6V0LGvqoo.wav"
    },
    "tangoflux": {
     "label": "TangoFlux",
     "path": "assets/audio/audiocaps/tangoflux/YXi6V0LGvqoo.wav"
    },
    "resonate": {
     "label": "Resonate",
     "path": "assets/audio/audiocaps/resonate/YXi6V0LGvqoo.wav"
    },
    "base": {
     "label": "Base Model",
     "path": "assets/audio/audiocaps/base/YXi6V0LGvqoo.wav"
    }
   }
  },
  {
   "id": "Yf2fSxfvmkZQ",
   "index": 12,
   "caption": "A man talking as a metal thumps followed by a power tool sanding as a metal pot clanks",
   "source": "AudioCaps test split",
   "audio": {
    "semantic": {
     "label": "SemanticAudio",
     "path": "assets/audio/audiocaps/semantic/Yf2fSxfvmkZQ.wav"
    },
    "tangoflux": {
     "label": "TangoFlux",
     "path": "assets/audio/audiocaps/tangoflux/Yf2fSxfvmkZQ.wav"
    },
    "resonate": {
     "label": "Resonate",
     "path": "assets/audio/audiocaps/resonate/Yf2fSxfvmkZQ.wav"
    },
    "base": {
     "label": "Base Model",
     "path": "assets/audio/audiocaps/base/Yf2fSxfvmkZQ.wav"
    }
   }
  },
  {
   "id": "YrUq4w4EUSWA",
   "index": 13,
   "caption": "An electronic device buzzing followed by paper tearing and then crumpling as a bell rings before a toilet flushes",
   "source": "AudioCaps test split",
   "audio": {
    "semantic": {
     "label": "SemanticAudio",
     "path": "assets/audio/audiocaps/semantic/YrUq4w4EUSWA.wav"
    },
    "tangoflux": {
     "label": "TangoFlux",
     "path": "assets/audio/audiocaps/tangoflux/YrUq4w4EUSWA.wav"
    },
    "resonate": {
     "label": "Resonate",
     "path": "assets/audio/audiocaps/resonate/YrUq4w4EUSWA.wav"
    },
    "base": {
     "label": "Base Model",
     "path": "assets/audio/audiocaps/base/YrUq4w4EUSWA.wav"
    }
   }
  },
  {
   "id": "YSmdj6JFB9MQ",
   "index": 14,
   "caption": "Music plays and an adult female speaks, a flush occurs and water splashes and gurgles, and a young female speaks",
   "source": "AudioCaps test split",
   "audio": {
    "semantic": {
     "label": "SemanticAudio",
     "path": "assets/audio/audiocaps/semantic/YSmdj6JFB9MQ.wav"
    },
    "tangoflux": {
     "label": "TangoFlux",
     "path": "assets/audio/audiocaps/tangoflux/YSmdj6JFB9MQ.wav"
    },
    "resonate": {
     "label": "Resonate",
     "path": "assets/audio/audiocaps/resonate/YSmdj6JFB9MQ.wav"
    },
    "base": {
     "label": "Base Model",
     "path": "assets/audio/audiocaps/base/YSmdj6JFB9MQ.wav"
    }
   }
  },
  {
   "id": "YPvWI4p74UOs",
   "index": 15,
   "caption": "A man laughing followed by another man yelling in the distance then wind blowing in a microphone as a motor hums and birds chirp in the background",
   "source": "AudioCaps test split",
   "audio": {
    "semantic": {
     "label": "SemanticAudio",
     "path": "assets/audio/audiocaps/semantic/YPvWI4p74UOs.wav"
    },
    "tangoflux": {
     "label": "TangoFlux",
     "path": "assets/audio/audiocaps/tangoflux/YPvWI4p74UOs.wav"
    },
    "resonate": {
     "label": "Resonate",
     "path": "assets/audio/audiocaps/resonate/YPvWI4p74UOs.wav"
    },
    "base": {
     "label": "Base Model",
     "path": "assets/audio/audiocaps/base/YPvWI4p74UOs.wav"
    }
   }
  }
 ],
 "ttabench": [
  {
   "id": "prompt_1373",
   "index": 1,
   "caption": "A sustained car horn blaring from gridlocked traffic.",
   "source": "TTABench",
   "audio": {
    "semantic": {
     "label": "SemanticAudio",
     "path": "assets/audio/ttabench/semantic/prompt_1373.wav"
    },
    "tangoflux": {
     "label": "TangoFlux",
     "path": "assets/audio/ttabench/tangoflux/prompt_1373.wav"
    },
    "resonate": {
     "label": "Resonate",
     "path": "assets/audio/ttabench/resonate/prompt_1373.wav"
    },
    "base": {
     "label": "Base Model",
     "path": "assets/audio/ttabench/base/prompt_1373.wav"
    }
   }
  },
  {
   "id": "prompt_0190",
   "index": 2,
   "caption": "A train is chugging along blows its whistle and steam escapes from its engine",
   "source": "TTABench",
   "audio": {
    "semantic": {
     "label": "SemanticAudio",
     "path": "assets/audio/ttabench/semantic/prompt_0190.wav"
    },
    "tangoflux": {
     "label": "TangoFlux",
     "path": "assets/audio/ttabench/tangoflux/prompt_0190.wav"
    },
    "resonate": {
     "label": "Resonate",
     "path": "assets/audio/ttabench/resonate/prompt_0190.wav"
    },
    "base": {
     "label": "Base Model",
     "path": "assets/audio/ttabench/base/prompt_0190.wav"
    }
   }
  },
  {
   "id": "prompt_0701",
   "index": 3,
   "caption": "A dog barks, prompting a baby to cry, then the owner's apologetic soothing.",
   "source": "TTABench",
   "audio": {
    "semantic": {
     "label": "SemanticAudio",
     "path": "assets/audio/ttabench/semantic/prompt_0701.wav"
    },
    "tangoflux": {
     "label": "TangoFlux",
     "path": "assets/audio/ttabench/tangoflux/prompt_0701.wav"
    },
    "resonate": {
     "label": "Resonate",
     "path": "assets/audio/ttabench/resonate/prompt_0701.wav"
    },
    "base": {
     "label": "Base Model",
     "path": "assets/audio/ttabench/base/prompt_0701.wav"
    }
   }
  },
  {
   "id": "prompt_0548",
   "index": 4,
   "caption": "A blender whirs briefly as glasses clink, the doorbell rings, someone answers it, and a baby coos in the background.",
   "source": "TTABench",
   "audio": {
    "semantic": {
     "label": "SemanticAudio",
     "path": "assets/audio/ttabench/semantic/prompt_0548.wav"
    },
    "tangoflux": {
     "label": "TangoFlux",
     "path": "assets/audio/ttabench/tangoflux/prompt_0548.wav"
    },
    "resonate": {
     "label": "Resonate",
     "path": "assets/audio/ttabench/resonate/prompt_0548.wav"
    },
    "base": {
     "label": "Base Model",
     "path": "assets/audio/ttabench/base/prompt_0548.wav"
    }
   }
  },
  {
   "id": "prompt_1299",
   "index": 5,
   "caption": "A motorcycle revs aggressively, a trash can clangs against a wall, and a dog barks persistently.",
   "source": "TTABench",
   "audio": {
    "semantic": {
     "label": "SemanticAudio",
     "path": "assets/audio/ttabench/semantic/prompt_1299.wav"
    },
    "tangoflux": {
     "label": "TangoFlux",
     "path": "assets/audio/ttabench/tangoflux/prompt_1299.wav"
    },
    "resonate": {
     "label": "Resonate",
     "path": "assets/audio/ttabench/resonate/prompt_1299.wav"
    },
    "base": {
     "label": "Base Model",
     "path": "assets/audio/ttabench/base/prompt_1299.wav"
    }
   }
  },
  {
   "id": "prompt_1397",
   "index": 6,
   "caption": "A dog's persistent barking erupts, followed by the jingle of its owner quickly shortening the leash.",
   "source": "TTABench",
   "audio": {
    "semantic": {
     "label": "SemanticAudio",
     "path": "assets/audio/ttabench/semantic/prompt_1397.wav"
    },
    "tangoflux": {
     "label": "TangoFlux",
     "path": "assets/audio/ttabench/tangoflux/prompt_1397.wav"
    },
    "resonate": {
     "label": "Resonate",
     "path": "assets/audio/ttabench/resonate/prompt_1397.wav"
    },
    "base": {
     "label": "Base Model",
     "path": "assets/audio/ttabench/base/prompt_1397.wav"
    }
   }
  },
  {
   "id": "prompt_1036",
   "index": 7,
   "caption": "Sheep bleating intermittently while wind rustles through the wheat stalks.",
   "source": "TTABench",
   "audio": {
    "semantic": {
     "label": "SemanticAudio",
     "path": "assets/audio/ttabench/semantic/prompt_1036.wav"
    },
    "tangoflux": {
     "label": "TangoFlux",
     "path": "assets/audio/ttabench/tangoflux/prompt_1036.wav"
    },
    "resonate": {
     "label": "Resonate",
     "path": "assets/audio/ttabench/resonate/prompt_1036.wav"
    },
    "base": {
     "label": "Base Model",
     "path": "assets/audio/ttabench/base/prompt_1036.wav"
    }
   }
  },
  {
   "id": "prompt_1445",
   "index": 8,
   "caption": "The driver argues over hands-free phone while tires hydroplane through water. A child's tablet plays cartoons and seat fabric squeaks.",
   "source": "TTABench",
   "audio": {
    "semantic": {
     "label": "SemanticAudio",
     "path": "assets/audio/ttabench/semantic/prompt_1445.wav"
    },
    "tangoflux": {
     "label": "TangoFlux",
     "path": "assets/audio/ttabench/tangoflux/prompt_1445.wav"
    },
    "resonate": {
     "label": "Resonate",
     "path": "assets/audio/ttabench/resonate/prompt_1445.wav"
    },
    "base": {
     "label": "Base Model",
     "path": "assets/audio/ttabench/base/prompt_1445.wav"
    }
   }
  },
  {
   "id": "prompt_1328",
   "index": 9,
   "caption": "Rain pelts windows while wipers struggle, heater blows air, passenger snores softly, and a phone alarm chirps repeatedly.",
   "source": "TTABench",
   "audio": {
    "semantic": {
     "label": "SemanticAudio",
     "path": "assets/audio/ttabench/semantic/prompt_1328.wav"
    },
    "tangoflux": {
     "label": "TangoFlux",
     "path": "assets/audio/ttabench/tangoflux/prompt_1328.wav"
    },
    "resonate": {
     "label": "Resonate",
     "path": "assets/audio/ttabench/resonate/prompt_1328.wav"
    },
    "base": {
     "label": "Base Model",
     "path": "assets/audio/ttabench/base/prompt_1328.wav"
    }
   }
  },
  {
   "id": "prompt_0584",
   "index": 10,
   "caption": "Simultaneous sounds of typing, paper rustling, clock ticking, and distant traffic create a work ambiance.",
   "source": "TTABench",
   "audio": {
    "semantic": {
     "label": "SemanticAudio",
     "path": "assets/audio/ttabench/semantic/prompt_0584.wav"
    },
    "tangoflux": {
     "label": "TangoFlux",
     "path": "assets/audio/ttabench/tangoflux/prompt_0584.wav"
    },
    "resonate": {
     "label": "Resonate",
     "path": "assets/audio/ttabench/resonate/prompt_0584.wav"
    },
    "base": {
     "label": "Base Model",
     "path": "assets/audio/ttabench/base/prompt_0584.wav"
    }
   }
  },
  {
   "id": "prompt_1338",
   "index": 11,
   "caption": "Cyclist's heavy breathing overlaps with the rustling of a windbreaker.",
   "source": "TTABench",
   "audio": {
    "semantic": {
     "label": "SemanticAudio",
     "path": "assets/audio/ttabench/semantic/prompt_1338.wav"
    },
    "tangoflux": {
     "label": "TangoFlux",
     "path": "assets/audio/ttabench/tangoflux/prompt_1338.wav"
    },
    "resonate": {
     "label": "Resonate",
     "path": "assets/audio/ttabench/resonate/prompt_1338.wav"
    },
    "base": {
     "label": "Base Model",
     "path": "assets/audio/ttabench/base/prompt_1338.wav"
    }
   }
  },
  {
   "id": "prompt_0506",
   "index": 12,
   "caption": "A hair dryer runs while the cabinet door opens, a comb is used, and someone taps the mirror.",
   "source": "TTABench",
   "audio": {
    "semantic": {
     "label": "SemanticAudio",
     "path": "assets/audio/ttabench/semantic/prompt_0506.wav"
    },
    "tangoflux": {
     "label": "TangoFlux",
     "path": "assets/audio/ttabench/tangoflux/prompt_0506.wav"
    },
    "resonate": {
     "label": "Resonate",
     "path": "assets/audio/ttabench/resonate/prompt_0506.wav"
    },
    "base": {
     "label": "Base Model",
     "path": "assets/audio/ttabench/base/prompt_0506.wav"
    }
   }
  },
  {
   "id": "prompt_1053",
   "index": 13,
   "caption": "A single drum maintaining a steady military cadence during a parade performance.",
   "source": "TTABench",
   "audio": {
    "semantic": {
     "label": "SemanticAudio",
     "path": "assets/audio/ttabench/semantic/prompt_1053.wav"
    },
    "tangoflux": {
     "label": "TangoFlux",
     "path": "assets/audio/ttabench/tangoflux/prompt_1053.wav"
    },
    "resonate": {
     "label": "Resonate",
     "path": "assets/audio/ttabench/resonate/prompt_1053.wav"
    },
    "base": {
     "label": "Base Model",
     "path": "assets/audio/ttabench/base/prompt_1053.wav"
    }
   }
  },
  {
   "id": "prompt_0704",
   "index": 14,
   "caption": "Overlapping sounds of bouncing basketball, sneaker squeaks, cheers, and a whistle blow.",
   "source": "TTABench",
   "audio": {
    "semantic": {
     "label": "SemanticAudio",
     "path": "assets/audio/ttabench/semantic/prompt_0704.wav"
    },
    "tangoflux": {
     "label": "TangoFlux",
     "path": "assets/audio/ttabench/tangoflux/prompt_0704.wav"
    },
    "resonate": {
     "label": "Resonate",
     "path": "assets/audio/ttabench/resonate/prompt_0704.wav"
    },
    "base": {
     "label": "Base Model",
     "path": "assets/audio/ttabench/base/prompt_0704.wav"
    }
   }
  },
  {
   "id": "prompt_0649",
   "index": 15,
   "caption": "Threat is shouted, furniture overturns, items crash, door slams, and car speeds away.",
   "source": "TTABench",
   "audio": {
    "semantic": {
     "label": "SemanticAudio",
     "path": "assets/audio/ttabench/semantic/prompt_0649.wav"
    },
    "tangoflux": {
     "label": "TangoFlux",
     "path": "assets/audio/ttabench/tangoflux/prompt_0649.wav"
    },
    "resonate": {
     "label": "Resonate",
     "path": "assets/audio/ttabench/resonate/prompt_0649.wav"
    },
    "base": {
     "label": "Base Model",
     "path": "assets/audio/ttabench/base/prompt_0649.wav"
    }
   }
  }
 ],
 "editing": [
  {
   "id": "YOmmPaIAXN0s_edit279",
   "audio_id": "YOmmPaIAXN0s",
   "index": 1,
   "edit_type": "subject_replace",
   "edit_type_raw": "word_replace",
   "source_caption": "A man speaks with some light cracking",
   "target_caption": "A woman speaks with some light cracking",
   "improvement_tar": 0.243,
   "clap_pred_tar": 0.259,
   "metadata_source": "flowedit_eval_results/minimal_edit_final/cfg5.0_20260521_002608/clap_edit_per_sample.json",
   "audio": {
    "source": {
     "label": "Source",
     "path": "assets/audio/editing/source/YOmmPaIAXN0s_edit279.wav"
    },
    "semantic": {
     "label": "SemanticAudio Edit",
     "path": "assets/audio/editing/semantic/YOmmPaIAXN0s_edit279.wav"
    }
   }
  },
  {
   "id": "YUCy1BEx8jBE_edit332",
   "audio_id": "YUCy1BEx8jBE",
   "index": 2,
   "edit_type": "subject_replace",
   "edit_type_raw": "word_replace",
   "source_caption": "A man talks while water burbles",
   "target_caption": "A woman talks while water burbles",
   "improvement_tar": 0.195,
   "clap_pred_tar": 0.435,
   "metadata_source": "flowedit_eval_results/minimal_edit_final/cfg5.0_20260521_002608/clap_edit_per_sample.json",
   "audio": {
    "source": {
     "label": "Source",
     "path": "assets/audio/editing/source/YUCy1BEx8jBE_edit332.wav"
    },
    "semantic": {
     "label": "SemanticAudio Edit",
     "path": "assets/audio/editing/semantic/YUCy1BEx8jBE_edit332.wav"
    }
   }
  },
  {
   "id": "YC8kR19NvynA_edit169",
   "audio_id": "YC8kR19NvynA",
   "index": 3,
   "edit_type": "subject_replace",
   "edit_type_raw": "word_replace",
   "source_caption": "A male speaking over a microphone",
   "target_caption": "A female speaking over a microphone",
   "improvement_tar": 0.143,
   "clap_pred_tar": 0.348,
   "metadata_source": "flowedit_eval_results/minimal_edit_final/cfg5.0_20260521_002608/clap_edit_per_sample.json",
   "audio": {
    "source": {
     "label": "Source",
     "path": "assets/audio/editing/source/YC8kR19NvynA_edit169.wav"
    },
    "semantic": {
     "label": "SemanticAudio Edit",
     "path": "assets/audio/editing/semantic/YC8kR19NvynA_edit169.wav"
    }
   }
  },
  {
   "id": "Y3wV3ST-c4PE_edit034",
   "audio_id": "Y3wV3ST-c4PE",
   "index": 4,
   "edit_type": "sound_addition",
   "edit_type_raw": "sound_addition",
   "source_caption": "Rhythmic ticking is occurring",
   "target_caption": "Rhythmic ticking is occurring with water flowing nearby",
   "improvement_tar": 0.179,
   "clap_pred_tar": 0.539,
   "metadata_source": "flowedit_eval_results/new_types/cfg5.0_20260521_110849/clap_edit_per_sample.json",
   "audio": {
    "source": {
     "label": "Source",
     "path": "assets/audio/editing/source/Y3wV3ST-c4PE_edit034.wav"
    },
    "semantic": {
     "label": "SemanticAudio Edit",
     "path": "assets/audio/editing/semantic/Y3wV3ST-c4PE_edit034.wav"
    }
   }
  },
  {
   "id": "Y13CBvjHZhOA_edit011",
   "audio_id": "Y13CBvjHZhOA",
   "index": 5,
   "edit_type": "sound_addition",
   "edit_type_raw": "sound_addition",
   "source_caption": "Engine noise with other engines passing by",
   "target_caption": "Engine noise with other engines passing by followed by a bell ringing",
   "improvement_tar": 0.204,
   "clap_pred_tar": 0.365,
   "metadata_source": "flowedit_eval_results/new_types/cfg5.0_20260521_110849/clap_edit_per_sample.json",
   "audio": {
    "source": {
     "label": "Source",
     "path": "assets/audio/editing/source/Y13CBvjHZhOA_edit011.wav"
    },
    "semantic": {
     "label": "SemanticAudio Edit",
     "path": "assets/audio/editing/semantic/Y13CBvjHZhOA_edit011.wav"
    }
   }
  },
  {
   "id": "Y1QNLMF-Kl_s_edit014",
   "audio_id": "Y1QNLMF-Kl_s",
   "index": 6,
   "edit_type": "sound_addition",
   "edit_type_raw": "sound_addition",
   "source_caption": "A woman delivering a speech",
   "target_caption": "A woman delivering a speech with birds chirping in the background",
   "improvement_tar": 0.148,
   "clap_pred_tar": 0.36,
   "metadata_source": "flowedit_eval_results/new_types/cfg5.0_20260521_110849/clap_edit_per_sample.json",
   "audio": {
    "source": {
     "label": "Source",
     "path": "assets/audio/editing/source/Y1QNLMF-Kl_s_edit014.wav"
    },
    "semantic": {
     "label": "SemanticAudio Edit",
     "path": "assets/audio/editing/semantic/Y1QNLMF-Kl_s_edit014.wav"
    }
   }
  },
  {
   "id": "YCMNlIW6Lkwc_edit090",
   "audio_id": "YCMNlIW6Lkwc",
   "index": 7,
   "edit_type": "sound_removal",
   "edit_type_raw": "sound_removal",
   "source_caption": "Camera muffling followed by a popping explosion as a group of people talk and silverware clacks in the background",
   "target_caption": "Camera muffling",
   "improvement_tar": 0.368,
   "clap_pred_tar": 0.397,
   "metadata_source": "flowedit_eval_results/new_types/cfg5.0_20260521_110849/clap_edit_per_sample.json",
   "audio": {
    "source": {
     "label": "Source",
     "path": "assets/audio/editing/source/YCMNlIW6Lkwc_edit090.wav"
    },
    "semantic": {
     "label": "SemanticAudio Edit",
     "path": "assets/audio/editing/semantic/YCMNlIW6Lkwc_edit090.wav"
    }
   }
  },
  {
   "id": "YBLMWD6fxhpo_edit088",
   "audio_id": "YBLMWD6fxhpo",
   "index": 8,
   "edit_type": "sound_removal",
   "edit_type_raw": "sound_removal",
   "source_caption": "A walking in the grass and a pig oink sound",
   "target_caption": "A walking in the grass",
   "improvement_tar": 0.352,
   "clap_pred_tar": 0.336,
   "metadata_source": "flowedit_eval_results/new_types/cfg5.0_20260521_110849/clap_edit_per_sample.json",
   "audio": {
    "source": {
     "label": "Source",
     "path": "assets/audio/editing/source/YBLMWD6fxhpo_edit088.wav"
    },
    "semantic": {
     "label": "SemanticAudio Edit",
     "path": "assets/audio/editing/semantic/YBLMWD6fxhpo_edit088.wav"
    }
   }
  },
  {
   "id": "Y9XqkKuTqEOM_edit078",
   "audio_id": "Y9XqkKuTqEOM",
   "index": 9,
   "edit_type": "sound_removal",
   "edit_type_raw": "sound_removal",
   "source_caption": "Wood brushing followed by a door creaking then wood clanking",
   "target_caption": "Wood brushing",
   "improvement_tar": 0.206,
   "clap_pred_tar": 0.309,
   "metadata_source": "flowedit_eval_results/new_types/cfg5.0_20260521_110849/clap_edit_per_sample.json",
   "audio": {
    "source": {
     "label": "Source",
     "path": "assets/audio/editing/source/Y9XqkKuTqEOM_edit078.wav"
    },
    "semantic": {
     "label": "SemanticAudio Edit",
     "path": "assets/audio/editing/semantic/Y9XqkKuTqEOM_edit078.wav"
    }
   }
  },
  {
   "id": "Yn4VktYihtJU_edit491",
   "audio_id": "Yn4VktYihtJU",
   "index": 10,
   "edit_type": "weather_change",
   "edit_type_raw": "word_replace",
   "source_caption": "Heavy wind followed by yelling and cheering",
   "target_caption": "Heavy rain followed by yelling and cheering",
   "improvement_tar": 0.18,
   "clap_pred_tar": 0.473,
   "metadata_source": "flowedit_eval_results/minimal_edit_final/cfg5.0_20260521_002608/clap_edit_per_sample.json",
   "audio": {
    "source": {
     "label": "Source",
     "path": "assets/audio/editing/source/Yn4VktYihtJU_edit491.wav"
    },
    "semantic": {
     "label": "SemanticAudio Edit",
     "path": "assets/audio/editing/semantic/Yn4VktYihtJU_edit491.wav"
    }
   }
  },
  {
   "id": "YC4JwGJQ2dUA_edit135",
   "audio_id": "YC4JwGJQ2dUA",
   "index": 11,
   "edit_type": "environment_add",
   "edit_type_raw": "environment_add",
   "source_caption": "Male yelling loudly and forcefully",
   "target_caption": "Male yelling loudly and forcefully in a large empty room",
   "improvement_tar": 0.065,
   "clap_pred_tar": 0.319,
   "metadata_source": "flowedit_eval_results/new_types/cfg5.0_20260521_110849/clap_edit_per_sample.json",
   "audio": {
    "source": {
     "label": "Source",
     "path": "assets/audio/editing/source/YC4JwGJQ2dUA_edit135.wav"
    },
    "semantic": {
     "label": "SemanticAudio Edit",
     "path": "assets/audio/editing/semantic/YC4JwGJQ2dUA_edit135.wav"
    }
   }
  },
  {
   "id": "Y8b9z7N25DmU_edit118",
   "audio_id": "Y8b9z7N25DmU",
   "index": 12,
   "edit_type": "environment_add",
   "edit_type_raw": "environment_add",
   "source_caption": "Bus coming to a stop and door opening",
   "target_caption": "Bus coming to a stop and door opening outdoors with wind blowing",
   "improvement_tar": 0.042,
   "clap_pred_tar": 0.414,
   "metadata_source": "flowedit_eval_results/new_types/cfg5.0_20260521_110849/clap_edit_per_sample.json",
   "audio": {
    "source": {
     "label": "Source",
     "path": "assets/audio/editing/source/Y8b9z7N25DmU_edit118.wav"
    },
    "semantic": {
     "label": "SemanticAudio Edit",
     "path": "assets/audio/editing/semantic/Y8b9z7N25DmU_edit118.wav"
    }
   }
  },
  {
   "id": "YHZ9O6sc7cLA_edit096",
   "audio_id": "YHZ9O6sc7cLA",
   "index": 13,
   "edit_type": "atmosphere_change",
   "edit_type_raw": "atmosphere_change",
   "source_caption": "A woman is speaking followed by a dog growling and barking",
   "target_caption": "A woman is speaking followed by a dog growling and barking echoing in a large space",
   "improvement_tar": 0.085,
   "clap_pred_tar": 0.486,
   "metadata_source": "flowedit_eval_results/new_types_part2/cfg5.0_20260521_114146/clap_edit_per_sample.json",
   "audio": {
    "source": {
     "label": "Source",
     "path": "assets/audio/editing/source/YHZ9O6sc7cLA_edit096.wav"
    },
    "semantic": {
     "label": "SemanticAudio Edit",
     "path": "assets/audio/editing/semantic/YHZ9O6sc7cLA_edit096.wav"
    }
   }
  },
  {
   "id": "YEUZaxaWqhwg_edit078",
   "audio_id": "YEUZaxaWqhwg",
   "index": 14,
   "edit_type": "atmosphere_change",
   "edit_type_raw": "atmosphere_change",
   "source_caption": "Wind blows and people speak with some rattling and hissing",
   "target_caption": "Wind blows and people speak with some rattling and hissing distorted and eerie",
   "improvement_tar": 0.088,
   "clap_pred_tar": 0.265,
   "metadata_source": "flowedit_eval_results/new_types_part2/cfg5.0_20260521_114146/clap_edit_per_sample.json",
   "audio": {
    "source": {
     "label": "Source",
     "path": "assets/audio/editing/source/YEUZaxaWqhwg_edit078.wav"
    },
    "semantic": {
     "label": "SemanticAudio Edit",
     "path": "assets/audio/editing/semantic/YEUZaxaWqhwg_edit078.wav"
    }
   }
  },
  {
   "id": "YmJ6ZO3xEcgw_edit058",
   "audio_id": "YmJ6ZO3xEcgw",
   "index": 15,
   "edit_type": "intensity_change",
   "edit_type_raw": "intensity_change",
   "source_caption": "A woman sneezing repeatedly",
   "target_caption": "A woman sneezing briefly",
   "improvement_tar": 0.076,
   "clap_pred_tar": 0.451,
   "metadata_source": "flowedit_eval_results/new_types_part2/cfg5.0_20260521_114146/clap_edit_per_sample.json",
   "audio": {
    "source": {
     "label": "Source",
     "path": "assets/audio/editing/source/YmJ6ZO3xEcgw_edit058.wav"
    },
    "semantic": {
     "label": "SemanticAudio Edit",
     "path": "assets/audio/editing/semantic/YmJ6ZO3xEcgw_edit058.wav"
    }
   }
  },
  {
   "id": "YGMP8m09j5vk_edit159",
   "audio_id": "YGMP8m09j5vk",
   "index": 16,
   "edit_type": "spatial_change",
   "edit_type_raw": "spatial_change",
   "source_caption": "Birds chirping in the foreground and background as a dog barks in the distance",
   "target_caption": "Birds chirping in the foreground and background as a dog barks nearby",
   "improvement_tar": 0.14,
   "clap_pred_tar": 0.39,
   "metadata_source": "flowedit_eval_results/new_types_part2/cfg5.0_20260521_114146/clap_edit_per_sample.json",
   "audio": {
    "source": {
     "label": "Source",
     "path": "assets/audio/editing/source/YGMP8m09j5vk_edit159.wav"
    },
    "semantic": {
     "label": "SemanticAudio Edit",
     "path": "assets/audio/editing/semantic/YGMP8m09j5vk_edit159.wav"
    }
   }
  },
  {
   "id": "YECw5Yf7QoMo_edit154",
   "audio_id": "YECw5Yf7QoMo",
   "index": 17,
   "edit_type": "spatial_change",
   "edit_type_raw": "spatial_change",
   "source_caption": "A man talking in the foreground as another man talks in the background followed by a crowd of people applauding",
   "target_caption": "A man talking in the foreground as another man talks in the foreground followed by a crowd of people applauding",
   "improvement_tar": 0.058,
   "clap_pred_tar": 0.419,
   "metadata_source": "flowedit_eval_results/new_types_part2/cfg5.0_20260521_114146/clap_edit_per_sample.json",
   "audio": {
    "source": {
     "label": "Source",
     "path": "assets/audio/editing/source/YECw5Yf7QoMo_edit154.wav"
    },
    "semantic": {
     "label": "SemanticAudio Edit",
     "path": "assets/audio/editing/semantic/YECw5Yf7QoMo_edit154.wav"
    }
   }
  },
  {
   "id": "Yi2yhbckq3p0_edit454",
   "audio_id": "Yi2yhbckq3p0",
   "index": 18,
   "edit_type": "vehicle_replace",
   "edit_type_raw": "word_replace",
   "source_caption": "A car horn honks once shortly and then two long honks while another car honks",
   "target_caption": "A truck horn honks once shortly and then two long honks while another car honks",
   "improvement_tar": 0.118,
   "clap_pred_tar": 0.38,
   "metadata_source": "flowedit_eval_results/minimal_edit_final/cfg5.0_20260521_002608/clap_edit_per_sample.json",
   "audio": {
    "source": {
     "label": "Source",
     "path": "assets/audio/editing/source/Yi2yhbckq3p0_edit454.wav"
    },
    "semantic": {
     "label": "SemanticAudio Edit",
     "path": "assets/audio/editing/semantic/Yi2yhbckq3p0_edit454.wav"
    }
   }
  },
  {
   "id": "YqZEIs6tS5vk_edit525",
   "audio_id": "YqZEIs6tS5vk",
   "index": 19,
   "edit_type": "vehicle_replace",
   "edit_type_raw": "word_replace",
   "source_caption": "A truck engine revving followed by tires skidding as a crowd of people talk in the background",
   "target_caption": "A car engine revving followed by tires skidding as a crowd of people talk in the background",
   "improvement_tar": 0.087,
   "clap_pred_tar": 0.437,
   "metadata_source": "flowedit_eval_results/minimal_edit_final/cfg5.0_20260521_002608/clap_edit_per_sample.json",
   "audio": {
    "source": {
     "label": "Source",
     "path": "assets/audio/editing/source/YqZEIs6tS5vk_edit525.wav"
    },
    "semantic": {
     "label": "SemanticAudio Edit",
     "path": "assets/audio/editing/semantic/YqZEIs6tS5vk_edit525.wav"
    }
   }
  }
 ],
 "metrics": {
  "audiocaps": {
   "SemanticAudio d=128": {
    "FD": 19.1,
    "IS": 9.13,
    "CLAP_L": 0.381,
    "MOS": 3.72
   },
   "TangoFlux": {
    "FD": 22.6,
    "IS": 12.05,
    "CLAP_L": 0.361,
    "MOS": 3.85
   },
   "Resonate": {
    "FD": 16.0,
    "IS": 11.04,
    "CLAP_L": 0.403,
    "MOS": 3.78
   },
   "Base Model": {
    "FD": 25.0,
    "IS": 7.11,
    "CLAP_L": 0.318,
    "MOS": 3.42
   }
  },
  "ttabench": {
   "SemanticAudio d=128": {
    "CLAP_M": 0.459,
    "AES_CU": 5.187,
    "AES_PQ": 5.936
   },
   "TangoFlux": {
    "CLAP_M": 0.464,
    "AES_CU": 5.006,
    "AES_PQ": 5.701
   },
   "Resonate": {
    "CLAP_M": 0.459,
    "AES_CU": 5.258,
    "AES_PQ": 5.942
   },
   "Base Model": {
    "CLAP_M": 0.426,
    "AES_CU": 5.119,
    "AES_PQ": 5.897
   }
  },
  "editing": {
   "Semantic FlowEdit": {
    "CLAP_L": 0.351,
    "Delta_CLAP_L": 0.094,
    "MOS": 3.71
   },
   "TangoFlux FlowEdit": {
    "CLAP_L": 0.339,
    "Delta_CLAP_L": 0.081,
    "MOS": 3.54
   },
   "Resonate FlowEdit": {
    "CLAP_L": 0.322,
    "Delta_CLAP_L": 0.065,
    "MOS": 3.48
   },
   "DAC FlowEdit": {
    "CLAP_L": 0.29,
    "Delta_CLAP_L": 0.054,
    "MOS": 3.32
   }
  }
 }
}