feat: audio modes and built-in presets

Add 6 configurable audio modes (podcast, briefing, lecture, storyteller,
study-buddy, calm) that control the persona and structure used when
rewriting summaries for spoken playback. Ship 7 built-in presets
(morning-brief, commute-catch-up, deep-study, exam-prep, bedtime-read,
story-mode, team-debrief) with curated audio/style/tone/voice combos.

Rename "profiles" to "presets" in all user-facing surfaces while keeping
--profile and /profile as hidden aliases for backward compatibility.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Author: asanmateu

.gitignore

@@ -6,3 +6,4 @@ coverage/
 .DS_Store
 .env
 .env.*
+claude_notes/

CHANGELOG.md

@@ -9,10 +9,16 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Added
 
+- Audio modes: six personas for spoken summaries — podcast, briefing, lecture, storyteller, study-buddy, calm
+- `--audio-mode` CLI flag and `tldr config set audio-mode` command
+- Built-in presets: morning-brief, commute-catch-up, deep-study, exam-prep, bedtime-read, story-mode, team-debrief
+- `--preset` CLI flag and `tldr preset` command (list, create, delete, use, edit)
 - Chat export: press `Ctrl+s` in chat view to save the conversation as `chat.md` in the session directory; auto-saves on every subsequent message
 
 ### Changed
 
+- Renamed "profiles" to "presets" in CLI and docs (`--profile` and `tldr profile` still work as aliases)
+- Audio rewriting uses template-driven prompts instead of hardcoded podcast persona
 - Chat view: role labels changed from "You"/"AI" to `›`/`◆` symbols, messages wrapped in bordered card, consistent spacing between messages
 
 ## [2.1.0] - 2026-02-21

CLAUDE.md

@@ -15,11 +15,15 @@ When cutting a release, collect all `[Unreleased]` entries into a new version he
 
 1. **Update CHANGELOG.md** — Add entries under `[Unreleased]` for any user-facing change (Added, Fixed, Changed, Removed)
 2. **Write tests** — New features and bug fixes must have corresponding tests in `src/__tests__/`. Do not test external libraries or vendor APIs.
-3. **Update documentation** — If behavior changes, update the relevant docs in `docs/` and `README.md`
+3. **Update documentation** — After each feature, review `docs/` and `README.md` to ensure they reflect the current state of the product. Keep docs concise and non-redundant: prefer linking to the relevant `docs/` page over duplicating content in the README. The README should be a clean overview with a quick-start, not a full reference. A demo GIF placeholder lives at the top of the README — do not remove it.
 4. **Clean up obsolete code** — Remove dead references, unused types, stale docs entries. Don't leave commented-out code or backward-compat shims for removed features.
 5. **Run checks before finishing**:
    - `bun run test` — all tests pass
    - `bun run typecheck` — no type errors
    - `bun run check` — lint and format pass
-6. Always keep tldr-desktop in sync, especially when inserting breaking changes and new features.
-7. Keep code clean, if possible better than you found it.
+6. **Test strategy**:
+   - **Unit tests**: Pure functions, constants, validation. Every new type/constant/exported function.
+   - **Integration tests**: Config resolution flows, save/load round-trips with temp filesystem.
+   - **App-level tests**: UI behavior via ink-testing-library — keybindings, state transitions, rendered output.
+7. Always keep tldr-desktop in sync, especially when inserting breaking changes and new features. After each session that introduces breaking changes, new features, or UI-affecting changes, add an entry to `claude_notes/desktop-sync-tracker.md` describing what tldr-desktop needs to catch up on.
+8. Keep code clean, if possible better than you found it.

README.md

@@ -22,9 +22,9 @@ I had a pile of to-reads growing at work and it was overwhelming — not because
 
 ## Audio
 
-Summaries aren't just read aloud. They're **rewritten as podcast-style scripts** adapted to your cognitive profile. ADHD mode leads with hooks. Dyslexia mode uses short punchy sentences. The result sounds like a brief podcast, not a screen reader.
+Summaries aren't just read aloud. They're **rewritten as spoken scripts** using a configurable audio mode — podcast, briefing, lecture, storyteller, study-buddy, or calm. Each mode has a distinct persona and delivery style. The script also adapts to your cognitive traits and tone. The result sounds natural, not robotic.
 
-Press **`a`** to listen. Press **`w`** to save the audio alongside your summary. Free with Edge TTS by default, or switch to OpenAI TTS for higher quality. See the [Audio guide](docs/audio.md) for voices, save-with-audio, and provider options.
+Press **`a`** to listen. Press **`w`** to save the audio alongside your summary. Use built-in presets like `morning-brief` or `bedtime-read` for curated listening experiences. Free with Edge TTS by default, or switch to OpenAI TTS for higher quality. See the [Audio guide](docs/audio.md) for modes, voices, and provider options.
 
 ---
 
@@ -62,10 +62,10 @@ Type `/` in interactive mode to access commands like `/history`, `/setup`, `/con
 
 1. You give it something to read (a URL, file, or text)
 2. It extracts the content automatically
-3. Claude generates a structured, learning-focused summary
+3. Your AI provider generates a structured, learning-focused summary
 4. The summary is saved to `~/Documents/tldr/`
 
-Summaries use short sentences, bullet points, bold key terms, and put the most important information first. This isn't just formatting — it's based on how people with dyslexia and ADHD process information.
+Summaries use short sentences, bullet points, bold key terms, and put the most important information first. This isn't just formatting — it's designed around cognitive accessibility research to work better for all types of readers.
 
 ---
 
@@ -81,7 +81,7 @@ tldr --model sonnet "https://..."    # Use a different model
 
 Choose from three color themes (coral, ocean, forest) and auto dark/light mode during setup, or change later with `tldr config set theme`.
 
-You can set up profiles for different contexts (work, study, casual) with different tones, styles, and cognitive trait settings.
+Use built-in presets (morning-brief, deep-study, bedtime-read, etc.) or create your own for different contexts with different tones, styles, audio modes, and cognitive trait settings.
 
 See the [Configuration guide](docs/configuration.md) for details.
 
@@ -99,7 +99,7 @@ Cognitive accessibility is the point, not a feature. Every summary is optimised
 | **ESL** | Common words only, jargon defined inline |
 | **Visual thinker** | Hierarchical layout, grouped ideas, numbered steps |
 
-Traits stack. Enable multiple with `tldr profile edit`.
+Traits stack. Enable multiple with `tldr preset edit`.
 
 ---
 
@@ -112,9 +112,9 @@ By default tldr uses Claude Code (`claude-code`). You can also use the Anthropic
 ## Documentation
 
 - [Installation](docs/installation.md) — Homebrew, standalone binaries, prerequisites
-- [Configuration](docs/configuration.md) — settings, profiles, tones, summary styles
+- [Configuration](docs/configuration.md) — settings, presets, tones, summary styles, audio modes
 - [Providers](docs/providers.md) — Claude Code, Anthropic API, Gemini, xAI, Ollama, Codex, and OpenAI-compatible setup
-- [Audio](docs/audio.md) — listening to summaries, voices, save with audio
+- [Audio](docs/audio.md) — audio modes, listening to summaries, voices, save with audio
 
 ---
 

docs/audio.md

@@ -1,16 +1,38 @@
 # Audio
 
-Summaries aren't just read aloud. They're **rewritten as podcast-style audio scripts** by the same AI provider you use for summarizing. The script adapts to your cognitive traits and tone setting. The result sounds like a brief podcast, not a screen reader.
+Summaries aren't just read aloud. They're **rewritten as audio scripts** by the same AI provider you use for summarizing, using a configurable **audio mode** that controls the persona and structure. The script adapts to your cognitive traits and tone setting. The result sounds like a natural spoken piece, not a screen reader.
 
 ## How audio works
 
 Three things happen when you press `a`:
 
 1. Your **summary** is sent to the AI provider (same one used for summarizing).
-2. The AI **rewrites it as a spoken script** — conversational, with hooks and natural transitions.
+2. The AI **rewrites it as a spoken script** using your chosen **audio mode** — each mode has a distinct persona, structure, and delivery style.
 3. The script is sent to a **TTS voice** (Edge TTS or OpenAI) for synthesis and playback.
 
-The rewrite adapts to your **cognitive traits**:
+### Audio modes
+
+Audio modes control how the script is structured and delivered:
+
+| Mode | Persona | Best for |
+|------|---------|----------|
+| **podcast** (default) | Conversational host with hooks and transitions | General listening |
+| **briefing** | Analyst delivering concise, numbered facts | Daily catch-ups |
+| **lecture** | Patient teacher building understanding progressively | Deep learning |
+| **storyteller** | Narrator weaving a compelling narrative | Story-driven content |
+| **study-buddy** | Study partner with quizzes and mnemonics | Exam prep, retention |
+| **calm** | Gentle, soothing narrator | Relaxed / bedtime listening |
+
+```bash
+tldr config set audio-mode briefing           # Change default
+tldr --audio-mode lecture "https://..."        # Override for one run
+```
+
+Built-in presets bundle an audio mode with matching style, tone, and voice settings — see [Presets](configuration.md#presets).
+
+### Cognitive traits
+
+The rewrite also adapts to your **cognitive traits**. These work with any audio mode:
 
 | Trait | How the audio script changes |
 |-------|------------------------------|
@@ -20,7 +42,7 @@ The rewrite adapts to your **cognitive traits**:
 | **ESL** | Common vocabulary. Specialized terms explained inline. |
 | **Visual thinker** | Spatial language. Word pictures. Narrative structure. |
 
-Your **tone** setting also shapes the script (casual, professional, academic, eli5).
+Traits stack — enable multiple with `tldr preset edit`. Your **tone** setting (casual, professional, academic, eli5) also shapes the script.
 
 This costs one extra API call, which is why it sounds natural instead of robotic.
 
@@ -63,7 +85,7 @@ tldr config set tts-provider openai
 tldr config set tts-provider edge-tts   # back to default
 ```
 
-You can also change the TTS provider in the profile editor (`tldr profile edit` / `/config`).
+You can also change the TTS provider in the preset editor (`tldr preset edit` / `/config`).
 
 ### TTS Model
 
@@ -117,8 +139,8 @@ When you switch TTS providers, the voice automatically resets to the new provide
 ## Speed, Pitch & Volume
 
 ```bash
-# Set via profile editor
-tldr profile edit
+# Set via preset editor
+tldr preset edit
 
 # Or via CLI
 tldr config set tts-speed 1.25        # 25% faster

docs/configuration.md

@@ -2,7 +2,7 @@
 
 Settings are stored at `~/.tldr/settings.json`. Run `tldr config` to see your current settings.
 
-Most settings are easier to change interactively with `tldr profile edit` or `/config` in interactive mode. The CLI commands below are for when you know exactly what you want.
+Most settings are easier to change interactively with `tldr preset edit` or `/config` in interactive mode. The CLI commands below are for when you know exactly what you want.
 
 ## Summary
 
@@ -43,13 +43,33 @@ tldr --style quick "https://example.com/article"
 
 | Key | Values | Example |
 |-----|--------|---------|
+| `audio-mode` | `podcast`, `briefing`, `lecture`, `storyteller`, `study-buddy`, `calm` | `tldr config set audio-mode briefing` |
 | `tts-provider` | `edge-tts`, `openai` | `tldr config set tts-provider openai` |
 | `tts-model` | Any OpenAI TTS model ID | `tldr config set tts-model tts-1-hd` |
 | `voice` | TTS voice name | `tldr config set voice en-US-GuyNeural` |
 | `tts-speed` | Positive number | `tldr config set tts-speed 1.25` |
 | `pitch` | `low`, `default`, `high` | `tldr config set pitch high` |
 | `volume` | `quiet`, `normal`, `loud` | `tldr config set volume loud` |
-> **Note:** Pitch and Volume only apply to Edge TTS. They are not supported by OpenAI TTS and will be hidden in the profile editor when OpenAI is selected.
+> **Note:** Pitch and Volume only apply to Edge TTS. They are not supported by OpenAI TTS and will be hidden in the preset editor when OpenAI is selected.
+
+### Audio Modes
+
+Audio modes control the persona and structure used when rewriting summaries for spoken playback.
+
+| Mode | Persona | Best for |
+|------|---------|----------|
+| **podcast** (default) | Conversational podcast host | General listening |
+| **briefing** | Concise analyst | Quick daily catch-ups |
+| **lecture** | Patient teacher | Deep learning sessions |
+| **storyteller** | Compelling narrator | Narrative-driven content |
+| **study-buddy** | Smart study partner | Exam prep and retention |
+| **calm** | Gentle, soothing narrator | Relaxed listening |
+
+Override for a single run:
+
+```bash
+tldr --audio-mode briefing "https://example.com/article"
+```
 
 See the [Audio guide](audio.md) for workflow details, voice personalities, and save-with-audio behavior.
 
@@ -66,30 +86,46 @@ Three appearance modes: **auto** (detects system setting, default), **dark**, **
 
 ## Global Settings
 
-These are top-level settings, not profile-specific. They are not available in the interactive profile editor.
+These are top-level settings, not preset-specific. They are not available in the interactive preset editor.
 
 | Key | Values | Example |
 |-----|--------|---------|
 | `apiKey` | Your API key | `tldr config set apiKey sk-ant-...` |
 | `baseUrl` | Custom API endpoint | `tldr config set baseUrl https://proxy.example.com` |
 | `maxTokens` | Number | `tldr config set maxTokens 2048` |
 | `output-dir` | Path | `tldr config set output-dir ~/summaries` |
-| `activeProfile` | Profile name | `tldr config set activeProfile work` |
+| `activeProfile` | Preset name | `tldr config set activeProfile work` |
+
+## Presets
+
+Presets let you save different settings for different contexts. tldr ships with 7 built-in presets you can use out of the box or clone and customise.
+
+### Built-in Presets
+
+| Preset | Audio Mode | Style | Tone | Speed | Voice |
+|--------|-----------|-------|------|-------|-------|
+| **morning-brief** | briefing | quick | professional | 1.15x | Guy |
+| **commute-catch-up** | podcast | standard | casual | 1.0x | Jenny |
+| **deep-study** | lecture | study-notes | academic | 0.9x | Aria |
+| **exam-prep** | study-buddy | study-notes | casual | 0.95x | Jenny |
+| **bedtime-read** | calm | standard | casual | 0.85x | Sonia |
+| **story-mode** | storyteller | detailed | casual | 1.0x | Guy |
+| **team-debrief** | briefing | standard | professional | 1.0x | Jenny |
 
-## Profiles
+Built-in presets are read-only. To customise one, create a preset with the same name — the built-in settings are used as a starting point.
 
-Profiles let you save different settings for different contexts (work, study, casual).
+### Managing Presets
 
 ```bash
-tldr profile list                 # List profiles (* = active)
-tldr profile create work          # Create a new profile
-tldr profile use work             # Switch active profile
-tldr profile edit work            # Edit profile settings (interactive)
-tldr profile delete work          # Delete a profile
-tldr --profile work <url>         # Use a profile for one run
+tldr preset list                  # List presets (* = active)
+tldr preset create work           # Create a new preset
+tldr preset use work              # Switch active preset
+tldr preset edit work             # Edit preset settings (interactive)
+tldr preset delete work           # Delete a preset
+tldr --preset morning-brief <url> # Use a preset for one run
 ```
 
-In interactive mode, type `/profile` to switch profiles without leaving the TUI.
+In interactive mode, type `/preset` to switch presets without leaving the TUI.
 
 ## Interactive Commands
 
@@ -99,9 +135,9 @@ In interactive mode, type `/` to access commands:
 |---------|-------------|
 | `/history` | Browse and resume past sessions |
 | `/setup` | Re-run the first-time setup wizard |
-| `/config` | Edit current profile settings |
+| `/config` | Edit current preset settings |
 | `/theme` | Change color theme |
-| `/profile` | Switch between profiles |
+| `/preset` | Switch between presets |
 | `/help` | Show shortcuts and commands |
 | `/quit` | Exit the app |
 
@@ -127,7 +163,7 @@ When no `styleModels` are configured, all styles default to Opus.
 
 ## Model Selection
 
-The interactive profile editor (`tldr profile edit` / `/config`) uses a free-text input for the model field. Type any model ID supported by your provider — for example `gpt-4o` for OpenAI, `gemini-2.5-flash` for Gemini, or `llama3.3` for Ollama.
+The interactive preset editor (`tldr preset edit` / `/config`) uses a free-text input for the model field. Type any model ID supported by your provider — for example `gpt-4o` for OpenAI, `gemini-2.5-flash` for Gemini, or `llama3.3` for Ollama.
 
 ### Anthropic Aliases
 
@@ -161,9 +197,9 @@ Environment variables override file settings:
 
 Settings are resolved in this order (first wins):
 
-1. CLI flags (`--model`, `--style`, `--profile`, `--provider`)
+1. CLI flags (`--model`, `--style`, `--preset`, `--provider`, `--audio-mode`)
 2. Environment variables
-3. Profile settings in `settings.json`
+3. Preset/profile settings in `settings.json`
 4. Built-in defaults
 
 ## Defaults
@@ -173,6 +209,7 @@ Settings are resolved in this order (first wins):
 | Provider | `claude-code` (Claude Code) |
 | Summary style | `standard` |
 | Model | Opus (`claude-opus-4-6`) |
+| Audio mode | `podcast` |
 | Cognitive traits | `dyslexia` |
 | Tone | `casual` |
 | Voice | `en-US-JennyNeural` |

src/App.tsx

@@ -325,6 +325,7 @@ export function App({ initialInput, showConfig, editProfile, overrides }: AppPro
           setConfigMode("edit");
           setState("config");
           break;
+        case "preset":
         case "profile":
           (async () => {
             const profileList = await listProfiles();

src/__tests__/app.test.tsx

@@ -139,6 +139,7 @@ const TEST_CONFIG: Config = {
   outputDir: "/tmp/tldr-test",
   ttsProvider: "edge-tts",
   ttsModel: "tts-1",
+  audioMode: "podcast" as const,
 };
 
 const TEST_EXTRACTION: ExtractionResult = {

src/__tests__/chat.test.ts

@@ -48,6 +48,7 @@ function makeTestConfig(overrides?: Partial<ResolvedConfig>): Config {
     outputDir: "/tmp/tldr-output",
     ttsProvider: "edge-tts" as const,
     ttsModel: "tts-1",
+    audioMode: "podcast" as const,
     ...overrides,
   };
 }

src/__tests__/commands.test.ts

@@ -42,10 +42,16 @@ describe("matchCommands", () => {
     expect(names).toContain("help");
   });
 
-  it("matches /profile command", () => {
+  it("matches /preset command (and /profile as alias)", () => {
     const result = matchCommands("/pr");
     expect(result).toHaveLength(1);
-    expect(result[0]?.name).toBe("profile");
+    expect(result[0]?.name).toBe("preset");
+  });
+
+  it("matches /profile alias to /preset", () => {
+    const result = matchCommands("/profile");
+    expect(result).toHaveLength(1);
+    expect(result[0]?.name).toBe("preset");
   });
 });