Initial commit: Nat20 Notes — TTRPG session transcription & summarization

frontend/src/pages/Diagnostics.tsx

@@ -0,0 +1,55 @@
+import { useEffect, useState } from "react";
+import { Link } from "react-router-dom";
+import { api } from "../api";
+
+const LABELS: Record<string, string> = {
+  ffmpeg: "ffmpeg",
+  gpu: "GPU / CUDA",
+  huggingface_token: "HuggingFace token",
+  llm_backend: "Summarization backend",
+  disk_space: "Disk space",
+};
+
+export default function Diagnostics() {
+  const [result, setResult] = useState<any>(null);
+  const [loading, setLoading] = useState(false);
+
+  const run = async () => {
+    setLoading(true);
+    const r = await api.getDiagnostics().catch((e) => ({ ok: false, checks: [], _error: e.message }));
+    setResult(r);
+    setLoading(false);
+  };
+
+  useEffect(() => {
+    run();
+  }, []);
+
+  return (
+    <div className="max-w-xl mx-auto py-16 px-4">
+      <Link to="/" className="text-sm text-ink/40 hover:text-brass">&larr; Back</Link>
+      <h1 className="font-display text-3xl mt-4 mb-2">System check</h1>
+      <p className="text-ink/60 mb-6">Run this before your first session, or any time something isn't working.</p>
+
+      <button className="btn-secondary mb-6" onClick={run} disabled={loading}>
+        {loading ? "Checking..." : "Run checks again"}
+      </button>
+
+      {result?._error && (
+        <div className="card p-4 text-ember">Couldn't reach the backend: {result._error}</div>
+      )}
+
+      <div className="space-y-3">
+        {result?.checks?.map((c: any) => (
+          <div key={c.name} className="card p-4 flex gap-3">
+            <span className={`w-2 h-2 rounded-full mt-1.5 shrink-0 ${c.ok ? "bg-moss" : "bg-ember"}`} />
+            <div>
+              <div className="font-medium">{LABELS[c.name] || c.name}</div>
+              <div className="text-sm text-ink/60 whitespace-pre-wrap">{c.message}</div>
+            </div>
+          </div>
+        ))}
+      </div>
+    </div>
+  );
+}

frontend/src/pages/Notes.tsx

@@ -0,0 +1,35 @@
+import { useEffect, useState } from "react";
+import { Link, useParams } from "react-router-dom";
+import { api } from "../api";
+
+export default function Notes() {
+  const { id } = useParams();
+  const [notes, setNotes] = useState<any>(null);
+  const [tab, setTab] = useState<"dm" | "player">("dm");
+
+  useEffect(() => {
+    api.getNotes(id!).then(setNotes);
+  }, [id]);
+
+  if (!notes) return null;
+
+  return (
+    <div className="max-w-3xl mx-auto py-16 px-4">
+      <Link to={`/sessions/${id}`} className="text-sm text-ink/40 hover:text-brass">&larr; Back</Link>
+      <h1 className="font-display text-3xl mt-4 mb-6">Session notes</h1>
+
+      <div className="flex gap-2 mb-6">
+        <button className={tab === "dm" ? "btn-primary" : "btn-secondary"} onClick={() => setTab("dm")}>
+          DM notes
+        </button>
+        <button className={tab === "player" ? "btn-primary" : "btn-secondary"} onClick={() => setTab("player")}>
+          Player recap
+        </button>
+      </div>
+
+      <div className="card p-6 whitespace-pre-wrap leading-relaxed">
+        {tab === "dm" ? notes.dm_notes : notes.player_recap}
+      </div>
+    </div>
+  );
+}

frontend/src/pages/SessionDetail.tsx

@@ -0,0 +1,257 @@
+import { useEffect, useState, useCallback, useRef } from 'react';
+import { Link, useNavigate, useParams } from 'react-router-dom';
+import { api, Job } from '../api';
+
+function Toast({ message, onDone }: { message: string; onDone: () => void }) {
+  useEffect(() => { const t = setTimeout(onDone, 2500); return () => clearTimeout(t); }, [onDone]);
+  return (
+    <div className="fixed top-4 left-1/2 -translate-x-1/2 z-50 bg-emerald-700 text-white px-5 py-3 rounded-xl shadow-2xl text-sm font-medium">
+      {message}
+    </div>
+  );
+}
+
+function formatDuration(seconds: number): string {
+  const h = Math.floor(seconds / 3600);
+  const m = Math.floor((seconds % 3600) / 60);
+  const s = Math.floor(seconds % 60);
+  if (h > 0) return `${h}:${m.toString().padStart(2, '0')}:${s.toString().padStart(2, '0')}`;
+  return `${m}:${s.toString().padStart(2, '0')}`;
+}
+
+function ElapsedTimer({ createdAt, endedAt }: { createdAt: number; endedAt?: number }) {
+  const [elapsed, setElapsed] = useState(0);
+  useEffect(() => {
+    if (endedAt !== undefined) {
+      setElapsed(Math.floor(endedAt - createdAt));
+      return;
+    }
+    const tick = () => setElapsed(Math.floor(Date.now() / 1000 - createdAt));
+    tick();
+    const id = setInterval(tick, 1000);
+    return () => clearInterval(id);
+  }, [createdAt, endedAt]);
+  return <span className="font-mono text-xs">{formatDuration(elapsed)}</span>;
+}
+
+const STATUS_LABEL: Record<string, string> = {
+  uploaded: 'Ready to transcribe',
+  transcribing: 'Transcribing...',
+  transcribed: 'Ready to name speakers',
+  complete: 'Notes ready',
+};
+
+function JobStatusCard({ job, session, onRefresh, onDeleted }: { job: Job; session?: any; onRefresh: () => void; onDeleted: (strategy: string) => void }) {
+  const [showDeleteModal, setShowDeleteModal] = useState(false);
+  const [busy, setBusy] = useState(false);
+
+  const handleCancel = async () => {
+    if (!confirm('Are you sure you want to halt this job midway?')) return;
+    setBusy(true);
+    await api.cancelJob(job.id);
+    onRefresh();
+    setBusy(false);
+  };
+
+  const executeDeletion = async (strategy: 'all' | 'artifacts_only' | 'none') => {
+    setBusy(true);
+    await api.deleteJob(job.id, strategy);
+    setShowDeleteModal(false);
+    setBusy(false);
+    onDeleted(strategy);
+  };
+
+  const badgeClass =
+    job.status === 'done'
+      ? 'bg-emerald-500/20 text-emerald-400 border border-emerald-500/30'
+      : job.status === 'error'
+      ? 'bg-rose-500/20 text-rose-400 border border-rose-500/30'
+      : 'bg-amber-500/20 text-amber-400 border border-amber-500/30 animate-pulse';
+
+  return (
+    <div className="card p-5">
+      <div className="flex items-center justify-between mb-3">
+        <h3 className="font-semibold text-lg">{job.job_type === 'transcribe' ? 'Transcription' : 'Note generation'}</h3>
+        <span className={`px-2.5 py-1 text-xs font-bold rounded-full uppercase ${badgeClass}`}>
+          {job.status === 'error' ? 'Failed' : job.status}
+        </span>
+      </div>
+
+      {job.status === 'error' ? (
+        <div className="bg-rose-950/40 border border-rose-900/50 text-rose-300 text-sm p-3 rounded mb-3 font-mono">
+          <strong>Error{job.error_stage ? ` [${job.error_stage}]` : ''}:</strong> {job.error || 'Unexpected error.'}
+        </div>
+      ) : (
+        <p className="text-sm text-ink/60 mb-3">{job.progress || 'Waiting...'}</p>
+      )}
+
+      {job.status !== 'queued' && job.created_at && (
+        <div className="flex items-center gap-1 text-ink/40 mb-3">
+          <svg className="w-3.5 h-3.5" fill="none" viewBox="0 0 24 24" stroke="currentColor" strokeWidth={2}><path strokeLinecap="round" strokeLinejoin="round" d="M12 8v4l3 3m6-3a9 9 0 11-18 0 9 9 0 0118 0z"/></svg>
+          <ElapsedTimer
+            createdAt={job.created_at}
+            endedAt={job.status === 'done' || job.status === 'error' ? job.updated_at : undefined}
+          />
+        </div>
+      )}
+
+      {session?.audio_duration && (
+        <div className="text-xs text-ink/40 mb-3 font-mono">
+          Audio {formatDuration(session.audio_duration)}
+          {session.word_count ? ` \u2022 ${session.word_count.toLocaleString()} words` : ''}
+          {session.language ? ` \u2022 ${session.language}` : ''}
+        </div>
+      )}
+
+      <div className="flex gap-3">
+        {(job.status === 'queued' || job.status === 'running') && (
+          <button disabled={busy} onClick={handleCancel} className="btn-secondary text-sm">
+            Cancel
+          </button>
+        )}
+        {job.status === 'error' && job.job_type === 'transcribe' && (
+          <button
+            disabled={busy}
+            onClick={async () => { setBusy(true); await api.retryTranscription(job.session_id); setBusy(false); onRefresh(); }}
+            className="btn-primary text-sm"
+          >
+            Restart transcription
+          </button>
+        )}
+        <button disabled={busy} onClick={() => setShowDeleteModal(true)} className="btn-secondary text-sm ml-auto">
+          Delete...
+        </button>
+      </div>
+
+      {showDeleteModal && (
+        <div className="fixed inset-0 bg-black/60 flex items-center justify-center p-4 z-50">
+          <div className="bg-panel border border-white/10 max-w-md w-full rounded-xl p-6 shadow-2xl">
+            <h4 className="text-lg font-bold mb-2">Clean up session data?</h4>
+            <p className="text-sm text-ink/60 mb-6 leading-relaxed">
+              Choose what to remove along with this job's tracking record.
+            </p>
+            <div className="space-y-2">
+              <button onClick={() => executeDeletion('all')} className="w-full text-left card p-3 text-sm">
+                🗑️ <strong>Wipe everything</strong>
+                <span className="block text-xs text-ink/40 mt-0.5">Session record, notes, transcript, and the source audio/video.</span>
+              </button>
+              <button onClick={() => executeDeletion('artifacts_only')} className="w-full text-left card p-3 text-sm">
+                📝 <strong>Keep source file</strong>
+                <span className="block text-xs text-ink/40 mt-0.5">Clears notes/transcript but keeps the original recording.</span>
+              </button>
+              <button onClick={() => executeDeletion('none')} className="w-full text-left card p-3 text-sm">
+                ❌ <strong>Just clear this job</strong>
+                <span className="block text-xs text-ink/40 mt-0.5">Only removes the job tracking row. No files touched.</span>
+              </button>
+            </div>
+            <div className="mt-5 flex justify-end">
+              <button onClick={() => setShowDeleteModal(false)} className="text-sm text-ink/50 hover:text-ink">
+                Cancel
+              </button>
+            </div>
+          </div>
+        </div>
+      )}
+    </div>
+  );
+}
+
+export default function SessionDetail() {
+  const { id } = useParams();
+  const nav = useNavigate();
+  const [session, setSession] = useState<any>(null);
+  const [job, setJob] = useState<Job | null>(null);
+  const [toast, setToast] = useState<string | null>(null);
+
+  const load = useCallback(() => {
+    if (!id) return;
+    api.getSession(id).then((s) => {
+      setSession(s);
+      setJob(s.latest_job || null);
+    });
+  }, [id]);
+
+  useEffect(load, [load]);
+
+  // Poll while a job is actively running so status/progress stay live.
+  useEffect(() => {
+    if (!job || (job.status !== 'queued' && job.status !== 'running')) return;
+    const interval = setInterval(load, 2500);
+    return () => clearInterval(interval);
+  }, [job, load]);
+
+  // Preload audio as soon as transcription finishes so it's buffered for the speakers page
+  useEffect(() => {
+    if (session?.status !== 'transcribed' || !id) return;
+    const link = document.createElement('link');
+    link.rel = 'preload';
+    link.as = 'audio';
+    link.href = api.audioUrl(id);
+    document.head.appendChild(link);
+    return () => link.remove();
+  }, [session?.status, id]);
+
+  // Auto-redirect to speakers page once transcription completes
+  const autoRedirected = useRef(false);
+  useEffect(() => {
+    if (session?.status === 'transcribed') {
+      if (!autoRedirected.current) {
+        autoRedirected.current = true;
+        nav(`/sessions/${id}/speakers`);
+      }
+    } else {
+      autoRedirected.current = false;
+    }
+  }, [session?.status, id, nav]);
+
+  const retryTranscription = async () => {
+    if (!id) return;
+    await api.retryTranscription(id);
+    load();
+  };
+
+  const generateNotes = async () => {
+    if (!id) return;
+    await api.generateNotes(id);
+    load();
+  };
+
+  if (!session) return null;
+
+  return (
+    <div className="max-w-2xl mx-auto py-16 px-4">
+      {toast && <Toast message={toast} onDone={() => setToast(null)} />}
+      <Link to="/" className="text-sm text-ink/40 hover:text-brass">&larr; Back</Link>
+      <h1 className="font-display text-3xl mt-4 mb-1">{session.name}</h1>
+      <p className="text-ink/40 font-mono text-xs mb-6">{session.original_filename}</p>
+      <p className="text-brass mb-8">{STATUS_LABEL[session.status] ?? session.status}</p>
+
+      {job && (
+        <div className="mb-8">
+          <JobStatusCard job={job} session={session} onRefresh={load} onDeleted={(strategy) => {
+            if (strategy === 'all') {
+              setToast('Session wiped');
+              setTimeout(() => nav('/'), 600);
+            } else if (strategy === 'artifacts_only') {
+              setToast('Transcript and notes cleared — source file kept');
+              load();
+            } else {
+              setToast('Job record removed');
+              load();
+            }
+          }} />
+        </div>
+      )}
+
+      {session.status === 'uploaded' && !job && (
+        <button className="btn-primary" onClick={retryTranscription}>Start transcription</button>
+      )}
+      {session.status === 'complete' && (
+        <div className="mt-4 flex gap-3">
+          <button className="btn-secondary" onClick={generateNotes}>Regenerate notes</button>
+          <Link to={`/sessions/${id}/notes`} className="btn-primary">View notes</Link>
+        </div>
+      )}
+    </div>
+  );
+}

frontend/src/pages/Sessions.tsx

@@ -0,0 +1,95 @@
+import { useEffect, useState, useCallback } from "react";
+import { Link, useNavigate } from "react-router-dom";
+import { api } from "../api";
+
+const STATUS_LABEL: Record<string, string> = {
+  uploaded: "Ready to transcribe",
+  transcribing: "Transcribing...",
+  transcribed: "Ready to name speakers",
+  complete: "Notes ready",
+};
+
+export default function Sessions() {
+  const [sessions, setSessions] = useState<any[]>([]);
+  const [name, setName] = useState("");
+  const [file, setFile] = useState<File | null>(null);
+  const [uploadProgress, setUploadProgress] = useState<number | null>(null);
+  const nav = useNavigate();
+
+  const refresh = useCallback(() => api.listSessions().then(setSessions), []);
+  useEffect(() => {
+    refresh();
+  }, [refresh]);
+
+  // Poll while any session is actively processing
+  useEffect(() => {
+    const hasActive = sessions.some(s => s.status === 'transcribing');
+    if (!hasActive) return;
+    const interval = setInterval(refresh, 2500);
+    return () => clearInterval(interval);
+  }, [sessions, refresh]);
+
+  const upload = async () => {
+    if (!file || !name) return;
+    setUploadProgress(0);
+    const { session_id } = await api.createSession(name, file, setUploadProgress);
+    setUploadProgress(null);
+    nav(`/sessions/${session_id}`);
+  };
+
+  return (
+    <div className="max-w-3xl mx-auto py-16 px-4">
+      <div className="eyebrow mb-2">Nat20 Notes</div>
+      <h1 className="font-display text-4xl mb-8">Your Transcriptions</h1>
+
+      <div className="card p-6 mb-8">
+        <h2 className="font-display text-xl mb-4">New transcription</h2>
+        <div className="flex flex-col gap-3">
+          <input
+            className="input"
+            placeholder="Session name, e.g. Session 1 — The Frozen Crypt"
+            value={name}
+            onChange={(e) => setName(e.target.value)}
+          />
+          <input
+            type="file"
+            accept="audio/*,video/*"
+            onChange={(e) => setFile(e.target.files?.[0] ?? null)}
+            className="text-sm text-ink/70"
+          />
+          <button className="btn-primary self-start" disabled={!file || !name || uploadProgress !== null} onClick={upload}>
+            {uploadProgress !== null ? `Uploading ${uploadProgress}%` : "Upload"}
+          </button>
+        </div>
+      </div>
+
+      <div className="space-y-3">
+        {sessions.length === 0 && (
+          <p className="text-ink/40">No sessions yet — upload a recording to get started.</p>
+        )}
+        {sessions.map((s) => {
+          const job = s.latest_job;
+          const progress = job?.progress;
+          const displayStatus = s.status === 'transcribing' && progress
+            ? progress
+            : (STATUS_LABEL[s.status] ?? s.status);
+          const statusColor = job?.status === 'error' ? 'text-rose-400' : 'text-brass';
+          return (
+            <Link key={s.id} to={`/sessions/${s.id}`} className="card p-4 flex items-center justify-between hover:border-brass/40 border border-transparent block">
+              <div>
+                <div className="font-medium">{s.name}</div>
+                <div className="text-xs text-ink/40 font-mono">{s.original_filename}</div>
+              </div>
+              <div className={"text-sm " + statusColor}>{displayStatus}</div>
+            </Link>
+          );
+        })}
+      </div>
+
+      <div className="mt-10 flex gap-4">
+        <Link to="/diagnostics" className="text-sm text-ink/40 hover:text-brass">System check</Link>
+        <Link to="/settings" className="text-sm text-ink/40 hover:text-brass">Settings</Link>
+      </div>
+    </div>
+  );
+}

frontend/src/pages/Settings.tsx

@@ -0,0 +1,68 @@
+import { useEffect, useState } from "react";
+import { Link } from "react-router-dom";
+import { api } from "../api";
+
+export default function Settings() {
+  const [settings, setSettings] = useState<any>(null);
+  const [saved, setSaved] = useState(false);
+
+  useEffect(() => {
+    api.getSettings().then(setSettings);
+  }, []);
+
+  if (!settings) return null;
+
+  const set = (k: string, v: any) => setSettings({ ...settings, [k]: v });
+
+  const save = async () => {
+    await api.updateSettings(settings);
+    setSaved(true);
+    setTimeout(() => setSaved(false), 2000);
+  };
+
+  return (
+    <div className="max-w-xl mx-auto py-16 px-4">
+      <Link to="/" className="text-sm text-ink/40 hover:text-brass">&larr; Back</Link>
+      <h1 className="font-display text-3xl mt-4 mb-8">Settings</h1>
+
+      <div className="card p-6 space-y-4 mb-6">
+        <h2 className="eyebrow">Transcription</h2>
+        <select className="input w-full" value={settings.whisper_model} onChange={(e) => set("whisper_model", e.target.value)}>
+          <option value="tiny">Tiny</option>
+          <option value="small">Small</option>
+          <option value="medium">Medium</option>
+          <option value="large-v3">Large v3</option>
+        </select>
+        <input className="input w-full" type="password" placeholder="HuggingFace token" value={settings.hf_token} onChange={(e) => set("hf_token", e.target.value)} />
+      </div>
+
+      <div className="card p-6 space-y-4 mb-6">
+        <h2 className="eyebrow">Summarization</h2>
+        <div className="flex gap-3">
+          <button className={settings.llm_mode === "ollama" ? "btn-primary" : "btn-secondary"} onClick={() => set("llm_mode", "ollama")}>Local (Ollama)</button>
+          <button className={settings.llm_mode === "api" ? "btn-primary" : "btn-secondary"} onClick={() => set("llm_mode", "api")}>Hosted API</button>
+        </div>
+        {settings.llm_mode === "ollama" ? (
+          <>
+            <input className="input w-full" value={settings.ollama_host} onChange={(e) => set("ollama_host", e.target.value)} />
+            <input className="input w-full" value={settings.ollama_model} onChange={(e) => set("ollama_model", e.target.value)} />
+          </>
+        ) : (
+          <>
+            <input className="input w-full" value={settings.api_base_url} onChange={(e) => set("api_base_url", e.target.value)} />
+            <input className="input w-full" type="password" value={settings.api_key} onChange={(e) => set("api_key", e.target.value)} />
+            <input className="input w-full" value={settings.api_model} onChange={(e) => set("api_model", e.target.value)} />
+          </>
+        )}
+      </div>
+
+      <div className="card p-6 space-y-3 mb-6">
+        <h2 className="eyebrow">Campaign context</h2>
+        <textarea className="input w-full h-32" value={settings.world_context} onChange={(e) => set("world_context", e.target.value)} placeholder="Paste world context directly..." />
+        <input className="input w-full" value={settings.world_context_path} onChange={(e) => set("world_context_path", e.target.value)} placeholder="...or path to a file inside the container" />
+      </div>
+
+      <button className="btn-primary" onClick={save}>{saved ? "Saved" : "Save settings"}</button>
+    </div>
+  );
+}

frontend/src/pages/Setup.tsx

@@ -0,0 +1,205 @@
+import { useState, useEffect } from 'react';
+import { api, AppSettings } from '../api';
+
+const DEFAULT_SETTINGS: AppSettings = {
+  onboarding_completed: false,
+  whisper_model: 'medium',
+  whisper_compute_type: 'int8',
+  hf_token: '',
+  llm_mode: 'ollama',
+  ollama_host: 'http://localhost:11434',
+  ollama_model: 'qwen2.5:7b',
+  api_base_url: 'https://api.openai.com/v1',
+  api_key: '',
+  api_model: 'gpt-4o-mini',
+  chunk_word_target: '2500',
+  world_context: '',
+  world_context_path: '',
+};
+
+export default function Setup({ onComplete }: { onComplete: () => void }) {
+  const [step, setStep] = useState(1);
+  const [settings, setSettings] = useState<AppSettings>(DEFAULT_SETTINGS);
+
+  useEffect(() => {
+    api.getSettings().then(data => {
+      if (data) {
+        setSettings({ ...DEFAULT_SETTINGS, ...data });
+      }
+    }).catch(err => {
+      console.error('Failed to connect to settings endpoint:', err);
+    });
+  }, []);
+
+  const handleNext = () => setStep(prev => prev + 1);
+  const handleBack = () => setStep(prev => prev - 1);
+
+  const handleSave = async () => {
+    await api.saveSettings({ ...settings, onboarding_completed: true });
+    onComplete();
+  };
+
+  return (
+    <div className="min-h-screen bg-slate-900 text-slate-100 flex items-center justify-center p-6">
+      <div className="w-full max-w-2xl bg-slate-800 border border-slate-700 rounded-xl p-8 shadow-2xl">
+        <div className="mb-8 flex justify-between items-center">
+          <h1 className="text-2xl font-bold tracking-tight text-teal-400">Nat20Notes Configuration Wizard</h1>
+          <span className="text-sm text-slate-400 font-mono">Step {step} of 2</span>
+        </div>
+
+        {step === 1 && (
+          <div className="space-y-6">
+            <div>
+              <h2 className="text-xl font-semibold mb-2">1. Audio & Diarization Engine</h2>
+              <p className="text-slate-400 text-sm mb-4">Set up your local Whisper sizing and secure deep speaker parsing models.</p>
+            </div>
+
+            <div>
+              <label className="block text-sm font-medium mb-2">Whisper Model Size</label>
+              <select
+                value={settings.whisper_model}
+                onChange={e => setSettings({ ...settings, whisper_model: e.target.value })}
+                className="w-full bg-slate-900 border border-slate-700 rounded px-3 py-2 text-white"
+              >
+                <option value="tiny">Tiny (~39M params)</option>
+                <option value="base">Base (~74M params)</option>
+                <option value="small">Small (~244M params)</option>
+                <option value="medium">Medium (~769M params)</option>
+                <option value="large-v3">Large V3 (~1.5B params - Recommended GPU)</option>
+              </select>
+            </div>
+
+            <div className="bg-slate-900 p-4 rounded border border-slate-700 space-y-3">
+              <label className="block text-sm font-medium">Hugging Face Access Token</label>
+              <input
+                type="password"
+                placeholder="hf_..."
+                value={settings.hf_token}
+                onChange={e => setSettings({ ...settings, hf_token: e.target.value })}
+                className="w-full bg-slate-800 border border-slate-700 rounded px-3 py-2 text-white font-mono"
+              />
+              <div className="text-xs text-slate-400 leading-relaxed">
+                💡 <strong>Diarization Requirements:</strong> To differentiate between multiple speakers, you must register a token:
+                <ol className="list-decimal list-inside ml-2 mt-1 space-y-1">
+                  <li>Create an account on <a href="https://huggingface.co" target="_blank" rel="noreferrer" className="text-teal-400 underline">huggingface.co</a></li>
+                  <li>Accept licensing conditions for <a href="https://huggingface.co/pyannote/speaker-diarization-3.1" target="_blank" rel="noreferrer" className="text-teal-400 underline">pyannote/speaker-diarization-3.1</a></li>
+                  <li>Accept licensing conditions for <a href="https://huggingface.co/pyannote/segmentation-3.0" target="_blank" rel="noreferrer" className="text-teal-400 underline">pyannote/segmentation-3.0</a></li>
+                  <li>Generate a <strong>Read</strong> token in Settings → Access Tokens.</li>
+                </ol>
+              </div>
+            </div>
+
+            <div className="flex justify-end pt-4">
+              <button onClick={handleNext} className="bg-teal-600 hover:bg-teal-500 text-white px-5 py-2 rounded font-medium">
+                Continue Next
+              </button>
+            </div>
+          </div>
+        )}
+
+        {step === 2 && (
+          <div className="space-y-6">
+            <div>
+              <h2 className="text-xl font-semibold mb-2">2. Summarization Engine (LLM)</h2>
+              <p className="text-slate-400 text-sm mb-4">Choose where your post-transcription formatting and smart summaries are generated.</p>
+            </div>
+
+            <div className="flex gap-4 border-b border-slate-700 pb-4">
+              {(['ollama', 'api'] as const).map(mode => (
+                <button
+                  key={mode}
+                  onClick={() => setSettings({ ...settings, llm_mode: mode })}
+                  className={`px-4 py-2 rounded uppercase text-xs font-bold tracking-wider ${settings.llm_mode === mode ? 'bg-teal-600 text-white' : 'bg-slate-900 text-slate-400'}`}
+                >
+                  {mode === 'ollama' ? 'Local (Ollama)' : 'Hosted API'}
+                </button>
+              ))}
+            </div>
+
+            {settings.llm_mode === 'ollama' ? (
+              <div className="space-y-4">
+                <div>
+                  <label className="block text-sm font-medium mb-1">Ollama Host Address</label>
+                  <input
+                    type="text"
+                    value={settings.ollama_host}
+                    onChange={e => setSettings({ ...settings, ollama_host: e.target.value })}
+                    className="w-full bg-slate-900 border border-slate-700 rounded px-3 py-2"
+                  />
+                </div>
+                <div>
+                  <label className="block text-sm font-medium mb-1">Model Name</label>
+                  <input
+                    type="text"
+                    placeholder="qwen2.5:7b"
+                    value={settings.ollama_model}
+                    onChange={e => setSettings({ ...settings, ollama_model: e.target.value })}
+                    className="w-full bg-slate-900 border border-slate-700 rounded px-3 py-2 font-mono"
+                  />
+                  <p className="text-xs text-slate-500 mt-1">Must match your local CLI entry for running "ollama pull &lt;model&gt;".</p>
+                </div>
+              </div>
+            ) : (
+              <div className="space-y-4">
+                <div>
+                  <label className="block text-sm font-medium mb-1">API Base URL</label>
+                  <input
+                    type="text"
+                    value={settings.api_base_url}
+                    onChange={e => setSettings({ ...settings, api_base_url: e.target.value })}
+                    className="w-full bg-slate-900 border border-slate-700 rounded px-3 py-2 font-mono"
+                  />
+                </div>
+                <div>
+                  <label className="block text-sm font-medium mb-1">API Key</label>
+                  <input
+                    type="password"
+                    value={settings.api_key}
+                    onChange={e => setSettings({ ...settings, api_key: e.target.value })}
+                    className="w-full bg-slate-900 border border-slate-700 rounded px-3 py-2 font-mono text-white"
+                    placeholder="sk-..."
+                  />
+                </div>
+                <div>
+                  <label className="block text-sm font-medium mb-1">Model Name</label>
+                  <input
+                    type="text"
+                    value={settings.api_model}
+                    onChange={e => setSettings({ ...settings, api_model: e.target.value })}
+                    className="w-full bg-slate-900 border border-slate-700 rounded px-3 py-2 font-mono"
+                  />
+                </div>
+              </div>
+            )}
+
+            <div className="border-t border-slate-700 pt-4 space-y-3">
+              <h3 className="text-sm font-semibold text-slate-300">Campaign context (optional)</h3>
+              <textarea
+                className="w-full bg-slate-900 border border-slate-700 rounded px-3 py-2 h-24"
+                placeholder="Paste world/campaign context directly..."
+                value={settings.world_context}
+                onChange={e => setSettings({ ...settings, world_context: e.target.value })}
+              />
+              <input
+                type="text"
+                className="w-full bg-slate-900 border border-slate-700 rounded px-3 py-2 font-mono text-sm"
+                placeholder="...or point to a local file"
+                value={settings.world_context_path}
+                onChange={e => setSettings({ ...settings, world_context_path: e.target.value })}
+              />
+            </div>
+
+            <div className="flex justify-between pt-4 border-t border-slate-700">
+              <button onClick={handleBack} className="bg-slate-700 hover:bg-slate-600 text-white px-5 py-2 rounded font-medium">
+                Back
+              </button>
+              <button onClick={handleSave} className="bg-emerald-600 hover:bg-emerald-500 text-white px-5 py-2 rounded font-medium">
+                Complete Setup & Launch
+              </button>
+            </div>
+          </div>
+        )}
+      </div>
+    </div>
+  );
+}

frontend/src/pages/Speakers.tsx

@@ -0,0 +1,181 @@
+import { useEffect, useRef, useState } from "react";
+import { Link, useNavigate, useParams } from "react-router-dom";
+import { api } from "../api";
+import { SessionReel, colorFor } from "../components/SessionReel";
+
+export default function Speakers() {
+  const { id } = useParams();
+  const nav = useNavigate();
+  const [speakers, setSpeakers] = useState<any[]>([]);
+  const [turns, setTurns] = useState<any[]>([]);
+  const [names, setNames] = useState<Record<string, string>>({});
+  const [currentTime, setCurrentTime] = useState(0);
+  const [duration, setDuration] = useState(0);
+  const [selected, setSelected] = useState<Set<string>>(new Set());
+  const [mergeTarget, setMergeTarget] = useState("");
+  const [busy, setBusy] = useState(false);
+  const [showConfirm, setShowConfirm] = useState(false);
+  const audioRef = useRef<HTMLAudioElement>(null);
+  const paletteRef = useRef(new Map<string, string>());
+
+  const load = () => {
+    api.getSpeakers(id!).then((sp) => {
+      setSpeakers(sp);
+      const n: Record<string, string> = {};
+      sp.forEach((s: any) => (n[s.raw_label] = s.display_name || ""));
+      setNames(n);
+    });
+    api.getTurns(id!).then(setTurns);
+  };
+  useEffect(load, [id]);
+
+  const save = async (raw_label: string) => {
+    await api.setSpeakerName(id!, raw_label, names[raw_label] || "");
+  };
+
+  const playAt = (t: number) => {
+    const audio = audioRef.current;
+    if (!audio) return;
+    if (audio.readyState < 2) return;
+    audio.pause();
+    audio.currentTime = t;
+    const p = audio.play();
+    if (p) p.catch(() => {});
+  };
+
+  const toggleSelected = (label: string) => {
+    const next = new Set(selected);
+    if (next.has(label)) next.delete(label); else next.add(label);
+    setSelected(next);
+  };
+
+  const doMerge = async () => {
+    if (!mergeTarget || selected.size < 2) return;
+    setBusy(true);
+    const sourceLabels = [...selected].filter(l => l !== mergeTarget);
+    await fetch(`/api/sessions/${id}/speakers/merge`, {
+      method: "POST",
+      headers: { "Content-Type": "application/json" },
+      body: JSON.stringify({ target_label: mergeTarget, source_labels: sourceLabels }),
+    });
+    setSelected(new Set());
+    setMergeTarget("");
+    setBusy(false);
+    load();
+  };
+
+  const handleDone = async () => {
+    setBusy(true);
+    await api.generateNotes(id!);
+    setBusy(false);
+    nav(`/sessions/${id}`);
+  };
+
+  const selectedArr = [...selected];
+
+  return (
+    <div className="max-w-4xl mx-auto py-16 px-4">
+      <Link to={`/sessions/${id}`} className="text-sm text-ink/40 hover:text-brass">&larr; Back</Link>
+      <h1 className="font-display text-3xl mt-4 mb-2">Name your speakers</h1>
+      <p className="text-ink/60 mb-6">Click anywhere on the reel to jump to that moment and hear who's talking.</p>
+
+      <audio
+        ref={audioRef}
+        src={api.audioUrl(id!)}
+        onTimeUpdate={(e) => setCurrentTime(e.currentTarget.currentTime)}
+        onLoadedMetadata={(e) => setDuration(e.currentTarget.duration)}
+        controls
+        className="w-full mb-3"
+      />
+
+      <div className="mb-8">
+        <SessionReel turns={turns} duration={duration} currentTime={currentTime} onSeek={playAt} />
+      </div>
+
+      {selected.size >= 2 && (
+        <div className="card p-4 mb-6 flex items-center gap-3 flex-wrap">
+          <span className="text-sm text-ink/60">{selected.size} selected</span>
+          <select
+            className="input text-sm flex-1 min-w-[200px]"
+            value={mergeTarget}
+            onChange={(e) => setMergeTarget(e.target.value)}
+          >
+            <option value="">Merge into...</option>
+            {selectedArr.map((l) => (
+              <option key={l} value={l}>{l} {names[l] ? `(${names[l]})` : ""}</option>
+            ))}
+          </select>
+          <button className="btn-primary text-sm" disabled={!mergeTarget || busy} onClick={doMerge}>
+            {busy ? "Merging..." : "Merge"}
+          </button>
+          <button className="btn-secondary text-sm" onClick={() => { setSelected(new Set()); setMergeTarget(""); }}>
+            Cancel
+          </button>
+        </div>
+      )}
+
+      <div className="grid gap-4 sm:grid-cols-2">
+        {speakers.map((sp) => (
+          <div key={sp.raw_label} className={"card p-4 " + (selected.has(sp.raw_label) ? "ring-2 ring-brass" : "")}>
+            <div className="flex items-center gap-2 mb-2">
+              <input
+                type="checkbox"
+                checked={selected.has(sp.raw_label)}
+                onChange={() => toggleSelected(sp.raw_label)}
+                className="accent-brass"
+              />
+              <span
+                className="w-3 h-3 rounded-full inline-block flex-shrink-0"
+                style={{ background: colorFor(sp.raw_label, paletteRef.current) }}
+              />
+              <span className="font-mono text-xs text-ink/40">{sp.raw_label}</span>
+            </div>
+            <input
+              className="input w-full mb-3"
+              placeholder="Player or character name"
+              value={names[sp.raw_label] || ""}
+              onChange={(e) => setNames({ ...names, [sp.raw_label]: e.target.value })}
+              onBlur={() => save(sp.raw_label)}
+            />
+            <div className="space-y-1">
+              {sp.samples.map((s: any, i: number) => (
+                <button
+                  key={i}
+                  onClick={() => playAt(s.start)}
+                  className="block text-left text-sm text-ink/60 hover:text-ink w-full truncate"
+                >
+                  ▶ {s.text}
+                </button>
+              ))}
+            </div>
+          </div>
+        ))}
+      </div>
+
+      <div className="mt-8">
+        <button className="btn-primary" onClick={() => setShowConfirm(true)}>
+          Done naming
+        </button>
+      </div>
+
+      {showConfirm && (
+        <div className="fixed inset-0 bg-black/60 flex items-center justify-center p-4 z-50">
+          <div className="bg-panel border border-white/10 max-w-md w-full rounded-xl p-6 shadow-2xl">
+            <h4 className="text-lg font-bold mb-2">Generate notes now?</h4>
+            <p className="text-sm text-ink/60 mb-6 leading-relaxed">
+              This will use your speaker names to create session notes. Are you sure?
+            </p>
+            <div className="flex gap-3 justify-end">
+              <button onClick={() => setShowConfirm(false)} className="btn-secondary">
+                Cancel
+              </button>
+              <button disabled={busy} onClick={handleDone} className="btn-primary">
+                {busy ? "Starting..." : "Yes, generate notes"}
+              </button>
+            </div>
+          </div>
+        </div>
+      )}
+    </div>
+  );
+}