{
  "schema_version": 1,
  "diagram_type": "sequence",
  "meta": {
    "title": "One dictation, from hotkey to paste",
    "output": "docs/diagrams/dictation-sequence.html",
    "locale": "en",
    "viewBox": [
      1080,
      696
    ],
    "column_fit": "spread",
    "animation": "trace",
    "quality_profile": "showcase",
    "repository": {
      "url": "https://github.com/mahfuzur/openvoicetype",
      "revision": "1fe87ef28b63b0fd0b7ab4f986b4b4026635f5d7"
    }
  },
  "participants": [
    {
      "id": "user",
      "type": "external",
      "label": "You",
      "sublabel": "hotkey and voice",
      "sources": [
        {
          "path": "app/Sources/VoiceToText/AppDelegate.swift",
          "line": 507,
          "end_line": 517,
          "label": "hotKeyPressed"
        }
      ]
    },
    {
      "id": "app",
      "type": "frontend",
      "label": "App",
      "sublabel": "Dictation · Recorder",
      "sources": [
        {
          "path": "app/Sources/VoiceToText/Dictation.swift",
          "line": 149,
          "end_line": 182,
          "label": "start"
        },
        {
          "path": "app/Sources/VoiceToText/Dictation.swift",
          "line": 386,
          "end_line": 399,
          "label": "prestart"
        },
        {
          "path": "app/Sources/VoiceToText/Dictation.swift",
          "line": 205,
          "end_line": 262,
          "label": "stop"
        }
      ]
    },
    {
      "id": "script",
      "type": "backend",
      "label": "dictate.sh",
      "sublabel": "transcribe · refine",
      "sources": [
        {
          "path": "scripts/dictate.sh",
          "line": 1219,
          "end_line": 1225,
          "label": "cmd_transcribe"
        },
        {
          "path": "scripts/dictate.sh",
          "line": 1242,
          "end_line": 1258,
          "label": "cmd_refine"
        },
        {
          "path": "scripts/dictate.sh",
          "line": 1125,
          "end_line": 1148,
          "label": "refine_text"
        }
      ]
    },
    {
      "id": "whisper",
      "type": "backend",
      "label": "whisper-server",
      "sublabel": "model kept loaded",
      "sources": [
        {
          "path": "scripts/dictate.sh",
          "line": 744,
          "end_line": 777,
          "label": "srv_start"
        },
        {
          "path": "scripts/dictate.sh",
          "line": 267,
          "end_line": 276,
          "label": "transcribe_server"
        }
      ]
    },
    {
      "id": "claude",
      "type": "cloud",
      "label": "claude -p",
      "sublabel": "your Claude Code",
      "sources": [
        {
          "path": "scripts/dictate.sh",
          "line": 579,
          "end_line": 602,
          "label": "claude_prestart"
        },
        {
          "path": "scripts/dictate.sh",
          "line": 609,
          "end_line": 641,
          "label": "claude_send"
        }
      ]
    },
    {
      "id": "target",
      "type": "external",
      "label": "Focused app",
      "sublabel": "where you type",
      "sources": [
        {
          "path": "app/Sources/VoiceToText/PasteTarget.swift",
          "line": 23,
          "end_line": 58,
          "label": "capture and check"
        },
        {
          "path": "app/Sources/VoiceToText/Paster.swift",
          "line": 13,
          "end_line": 61,
          "label": "paste"
        }
      ]
    }
  ],
  "segments": [
    {
      "from": 172,
      "to": 296,
      "label": "Press: the slow parts start while you speak"
    },
    {
      "from": 320,
      "to": 418,
      "label": "Transcribe"
    },
    {
      "from": 442,
      "to": 596,
      "label": "Clean up and paste"
    }
  ],
  "messages": [
    {
      "id": "press",
      "from": "user",
      "to": "app",
      "y": 184,
      "label": "press: pin app and window",
      "variant": "emphasis"
    },
    {
      "id": "launch",
      "from": "app",
      "to": "script",
      "y": 208,
      "label": "whisper-server start; refine",
      "variant": "emphasis"
    },
    {
      "id": "load-model",
      "from": "script",
      "to": "whisper",
      "y": 232,
      "label": "load model (0.6 s)",
      "variant": "dashed"
    },
    {
      "id": "prestart",
      "from": "script",
      "to": "claude",
      "y": 260,
      "label": "start claude -p early",
      "variant": "dashed"
    },
    {
      "id": "speak",
      "from": "user",
      "to": "app",
      "y": 284,
      "label": "speak, then press again",
      "variant": "emphasis"
    },
    {
      "id": "transcribe",
      "from": "app",
      "to": "script",
      "y": 332,
      "label": "transcribe <wav>"
    },
    {
      "id": "inference",
      "from": "script",
      "to": "whisper",
      "y": 356,
      "label": "POST /inference + prompt",
      "variant": "emphasis"
    },
    {
      "id": "whisper-text",
      "from": "whisper",
      "to": "script",
      "y": 384,
      "label": "text",
      "variant": "return"
    },
    {
      "id": "raw",
      "from": "script",
      "to": "app",
      "y": 408,
      "label": "raw transcript",
      "variant": "return"
    },
    {
      "id": "send",
      "from": "app",
      "to": "script",
      "y": 454,
      "label": "transcript on stdin",
      "variant": "emphasis"
    },
    {
      "id": "user-message",
      "from": "script",
      "to": "claude",
      "y": 478,
      "label": "one user message",
      "variant": "emphasis"
    },
    {
      "id": "result",
      "from": "claude",
      "to": "script",
      "y": 506,
      "label": "result (about 1 s)",
      "variant": "return"
    },
    {
      "id": "cleaned",
      "from": "script",
      "to": "app",
      "y": 530,
      "label": "guarded text + result file",
      "variant": "return"
    },
    {
      "id": "paste",
      "from": "app",
      "to": "target",
      "y": 558,
      "label": "same app and window? ⌘V",
      "variant": "security"
    },
    {
      "id": "read-paste",
      "from": "target",
      "to": "app",
      "y": 586,
      "label": "reads it; clipboard restored",
      "variant": "return"
    }
  ],
  "activations": [
    {
      "participant": "app",
      "from": 180,
      "to": 592,
      "type": "frontend"
    },
    {
      "participant": "script",
      "from": 204,
      "to": 536,
      "type": "backend"
    },
    {
      "participant": "whisper",
      "from": 228,
      "to": 238,
      "type": "backend"
    },
    {
      "participant": "whisper",
      "from": 352,
      "to": 390,
      "type": "backend"
    },
    {
      "participant": "claude",
      "from": 256,
      "to": 512,
      "type": "cloud"
    },
    {
      "participant": "target",
      "from": 554,
      "to": 592,
      "type": "external"
    }
  ],
  "cards": [
    {
      "dot": "emerald",
      "title": "Why it's fast",
      "items": [
        "Whisper and Claude start while you speak, not after",
        "One Claude process per dictation, so no earlier transcript stays in its context",
        "Stop to pasted: about 2.7 s for 8 s of speech with Claude"
      ]
    },
    {
      "dot": "orange",
      "title": "When a step fails",
      "items": [
        "whisper-server down: whisper-cli transcribes instead",
        "Claude fails or times out: S1-mini, then Whisper's own text",
        "No speech or Esc: the waiting refine gets empty input and exits"
      ]
    },
    {
      "dot": "rose",
      "title": "Before the paste",
      "items": [
        "The meaning guard keeps every number and \"not\", or uses Whisper's text",
        "A changed app, window or title means the text is copied instead",
        "Your clipboard comes back once the app has read the paste"
      ]
    }
  ]
}