com.octavius.audio.describe

{
  "$type": "com.atproto.lexicon.schema",
  "defs": {
    "audioDescription": {
      "properties": {
        "features": {
          "ref": "#audioFeatures",
          "type": "ref"
        },
        "prose": {
          "description": "A human-readable multi-line description of the audio.",
          "type": "string"
        },
        "service": {
          "const": "com.octavius.audio.describe",
          "type": "string"
        },
        "version": {
          "description": "Ears service version.",
          "type": "string"
        }
      },
      "required": [
        "service",
        "version",
        "features",
        "prose"
      ],
      "type": "object"
    },
    "audioFeatures": {
      "properties": {
        "audio": {
          "properties": {
            "channels": {
              "type": "integer"
            },
            "duration_seconds": {
              "type": "number"
            },
            "sample_rate": {
              "type": "integer"
            }
          },
          "type": "object"
        },
        "dynamics": {
          "properties": {
            "onset_density_per_second": {
              "type": "number"
            },
            "rms_dynamic_range_db": {
              "type": "number"
            }
          },
          "type": "object"
        },
        "key": {
          "properties": {
            "confidence": {
              "type": "number"
            },
            "mode": {
              "type": "string"
            },
            "name": {
              "type": "string"
            },
            "tonic": {
              "type": "string"
            }
          },
          "type": "object"
        },
        "tempo": {
          "properties": {
            "avg_beat_interval_seconds": {
              "type": "number"
            },
            "beat_count": {
              "type": "integer"
            },
            "bpm": {
              "type": "number"
            },
            "half_time_fold_applied": {
              "type": "boolean"
            }
          },
          "type": "object"
        },
        "texture": {
          "properties": {
            "percussive_share": {
              "type": "number"
            },
            "tonal_share": {
              "type": "number"
            }
          },
          "type": "object"
        },
        "timbre": {
          "properties": {
            "mfcc_mean": {
              "items": {
                "type": "number"
              },
              "type": "array"
            },
            "spectral_centroid_hz": {
              "type": "number"
            },
            "spectral_flatness": {
              "type": "number"
            },
            "spectral_flux": {
              "type": "number"
            },
            "spectral_rolloff_hz": {
              "type": "number"
            },
            "zero_crossing_rate": {
              "type": "number"
            }
          },
          "type": "object"
        }
      },
      "type": "object"
    },
    "main": {
      "description": "Analyze audio and describe it in structured text.",
      "input": {
        "encoding": "application/json",
        "schema": {
          "properties": {
            "url": {
              "description": "A URL to an audio file to analyze.",
              "format": "uri",
              "type": "string"
            }
          },
          "required": [],
          "type": "object"
        }
      },
      "output": {
        "encoding": "application/json",
        "schema": {
          "ref": "#audioDescription",
          "type": "ref"
        }
      },
      "type": "procedure"
    }
  },
  "description": "Analyze an audio file by URL and return a structured text description (key, tempo, timbre, dynamics). Backed by the alchemicAV Ears HTTP service; agents submit an audio URL and receive the description as text they can reason over.",
  "id": "com.octavius.audio.describe",
  "lexicon": 1
}