The Forge is a creative studio for songwriters, rappers, singers and poets. It runs entirely in your browser — no installation needed. It includes a block editor for writing lyrics (22 block types including verses, choruses, bridges, etc.), Pitch Engine for real-time voice analysis and training, a rhyme algorithm for Czech and English, a database of 107+ writing techniques, 30+ genre templates, a metronome, Notebook mode for free-form writing, and much more.

Yes, The Forge is completely free. No subscription, no paywall, no registration required. All features are available without limitations. If the app brings you value, you can voluntarily contribute via Buy Me a Coffee or PayPal — links are in the page footer. Contributions are not required and do not unlock any premium content.

No. The Forge is proprietary software. All rights reserved. The source code is not publicly accessible. Some components (rhyme algorithms, rhyme scheme detector, lyric engines) are protected as trade secrets. The license prohibits copying, modification, distribution and reverse engineering without the author's written consent.

The Forge was created by Lukáš Pištěk, an independent developer and musician. The app started as a tool for his own lyric writing and gradually grew into a comprehensive studio for the global songwriter community.

The app runs on Next.js 16 (App Router) + TypeScript 5 + React 19. Styling uses a custom CSS engine + Tailwind CSS 4. Audio processing uses the Web Audio API. Data is stored in localStorage and IndexedDB. Hosting is provided by Vercel, CDN and security by Cloudflare. Performance stack: • Wasm SIMD128 — YIN pitch detection in C++ compiled to Wasm (8 KB, 10× SIMD instructions), 2–4× faster than JS • Web Worker Pool — custom implementation with 3 strategies (persistent, run‑or‑drop, one‑shot), fallback chain • Benchmark — 8 tests of real algorithms, auto profile selection • WebGPU / OffscreenCanvas — detected in benchmark, not yet consumed

It fully supports Czech and English. All application text, technique databases, genre templates and inspiration templates are bilingual. Language is auto-detected from the browser and can be switched manually. German, French, Spanish and Polish are planned for the future.

No. The Forge works without registration. All your data stays locally in your browser. You create a profile by completing the onboarding — that is also local. Google login is optional and only serves for backup sync to Google Drive. Dropbox connection is also optional and works via OAuth (PKCE).

All your data (lyrics, projects, profile, recordings) is stored locally in your browser (localStorage and IndexedDB). Nothing is automatically sent to a server.

You have two backup options: Cloud (optional): Connect Dropbox or Google Drive via OAuth. The Forge then automatically saves your profile and media to the cloud. Sync happens automatically after changes. Local backup: You can download a ZIP file with all data (profile + projects + media). This file can be encrypted (AES-256-GCM). Emergency backup: If cloud sync fails, an emergency .forge file is created automatically.

You will lose all data — lyrics, projects, profile, settings. We strongly recommend creating a backup (ZIP export in profile) or connecting cloud sync before clearing your browser. The Forge has no server-side database to store your data.

Not directly. Data is stored locally in the browser. You can manually export your profile to ZIP (in profile settings) and import it on another device. Cloud sync (Dropbox/Google Drive) is for backup purposes, not cross-device synchronization — that is not yet implemented.

Both work similarly — storing profile and media to the cloud. Which one you choose depends on your preference — both services are fully supported.

Three writing modes: Block Editor: The main mode. Compose a song from blocks — each block has a type (VERSE, PRE-CHORUS, CHORUS, BRIDGE, SOLO, OUTRO, INTRO, FREE), a color and your text. Blocks can be dragged and reordered. Each mode has an (i) info button with help. Beginner Editor: Simplified single-block mode for newcomers. Advanced features (Pitch Engine, export, stress panel, rhyme scheme) are hidden until the user gains more experience. Notebook Mode: Free-form canvas for brainstorming — text blocks, sticky notes with photo/audio/rotation, label-free pins, highlighter (3 widths/6 colors), connection arrows, eraser, scissors, trash. Save full state or just text.

The Forge has two separate rule-based rhyme engines — no AI: Czech (387k words): G2P (grapheme-to-phoneme) phonetic analysis in 7 steps: normalization → grapheme splitting by sonority hierarchy → voicing assimilation → /v/ assimilation → final devoicing → /re:/ handling → mora assignment. The dictionary uses a trie structure (9.7 MB, lazy-loaded). Supports non-standard Czech (~70 slang entries). English (154k words + 11k IPA): Word list of 154k words + IPA phonetic dictionary (11k) + 185+ suffix patterns for out-of-dictionary words. Compounds are reconstructed. Scoring (0-100) considers 10 factors: zone match, vowel match (VOWEL_DISTANCE matrix), consonant group match (6 groups), POS quality, stress position, within-group consonant similarity, vowel harmony, length similarity, perfect match bonus, and structural penalty.

A comprehensive database of 107+ writing techniques in 6 categories: rhyme, rhythm, structure, delivery, repetition, and imagery. Each technique has bilingual content (CS/EN) including name, description, how-to, examples and tips. It also includes 53+ randomly generated motivational texts and 50 bilingual writing prompts (Today's Spark).

48+ templates in 13 categories: Pop & Dance: Pop, Synthpop, Dance-Pop, K-Pop Hip-Hop & Rap: Hip-Hop, Rap, Trap, Drill, Boom Bap, Grime, Phonk R&B & Soul: R&B Classic, R&B Fusion, Soul, Neo-Soul, Funk Rock: Rock, Alternative Rock, Hard Rock, Indie Rock, Grunge Metal & Hardcore: Metal, Metalcore, Thrash Metal Punk & Alternative: Punk, Pop Punk, Post-Punk, Indie, Emo Electronic & Dance: EDM, House, Techno, Drum & Bass Jazz & Blues: Jazz, Blues Folk & Country: Folk, Country, Ballad Classical & Orchestral: Classical, Cinematic Latin & Caribbean: Reggaeton, Dancehall, Reggae World & Folk Traditions: Afrobeat, Amapiano Special: Spoken Word, Musical Theatre, Experimental, Industrial, Custom Each template includes BPM range, default BPM, block structure (up to 14 block types), description, writing tips and things to avoid. Custom templates can be created with custom color and icon.

Yes, it offers several tools: Inspiration Panel: 53+ randomly generated motivational texts and quotes Today's Spark: 50 bilingual writing prompts on the home page, rotated daily Knowledge Book: 107+ techniques with examples and exercises Mood Matcher: Pick your mood (600+ phrases, 10 categories) and the app adjusts the visual theme Notebook Mode: Free-form canvas for brainstorming Word Bank: Personal word collection in localStorage with categories Focus Mode: (Ctrl+Shift+F) distraction-free environment

Yes. The editor can detect in real-time: alliteration, anaphora, internal rhyme, enjambment, colloquial Czech, and gibberish. It also detects rhyme schemes (AABB, ABAB, ABBA, ABCB, AAAA, AABA, AAAB) and counts syllables (Czech: vowel/diphthong counting, English: vowel groups + 18 special suffix rules).

Pitch Engine is a comprehensive real-time voice analysis and training tool. It runs entirely client-side (in the browser) using the Web Audio API — nothing is sent to a server. It detects pitch using the YIN algorithm (time-domain autocorrelation with parabolic interpolation, range 65 Hz–2000 Hz, polling every 50 ms). Optional pYIN (probabilistic YIN with Viterbi HMM) is available for multi-candidate detection.

Six training modes: Scales: Major and minor scales with real-time visual feedback on the musical staff. Intervals: 5 random intervals to sing. Chords: 3 chords (major/minor/dim/aug) with individual note detection. Coloraturas: Rapid scale runs up and down. Range Test: 12-second measurement with percentile-based voice classification. Warm-up: 4 exercise types: humming, breath, slides, volume.

Live metrics calculated every 200ms: Stability (0-100): MIDI variance — how steadily you hold a note. Vibrato: Detrended cents, zero-crossing rate (valid 3.5-7.5 Hz), rate + depth. Breath: RMS variance + level + voiced ratio. Resonance: Harmonic energy ratio (f0-8f0 vs total).

Yes, it auto-detects 6 voice types: soprano (C4-C6), mezzo-soprano (A3-A5), alto (F3-F5), tenor (C3-C5), baritone (G2-G4), bass (E2-E4). It also detects voice quality (breathy/trained) and analyzes tessitura (25th-75th percentile).

Yes. Pitch Engine can record using the MediaRecorder API (opus, 100ms timeslices) along with detected notes and metrics. Recordings are stored in IndexedDB and can be played back (RAF tick loop, 6s visible window). You can export them as MusicXML (for notation software), JSON or WebM audio.

Yes. Pitch Engine requires microphone access via getUserMedia. Without a microphone, analysis and training are not possible. The app will request permission on first launch of Pitch Engine. Voice detection uses Voice Activity Detection with a noise gate (envelope follower: 5ms attack, 100ms release).

Creators Wall is a community space for sharing song lyrics, collaboration, and feedback. You can publish your lyrics, discuss writing techniques, rate others' posts, and gain inspiration. All posts go through an approval process.

Click the "New Post" button (requires Google login). Go through the form: 1. Basic info: title (max 200 chars), post type (14 types in 5 groups — lyrics, collaboration, feedback, discussion, other), category (changes by type — genres, topics, or collab options), the text itself (min 20 chars) with Markdown editor formatting, and optional description. 2. Settings: display mode (nickname or anonymous), PIN for anonymous posts (4 digits for support), enable comments, copyright type (Copyright Free / Copyright), media links (images, YouTube, Spotify, Suno, Udio, SoundCloud), duration (1/7/31/90 days or forever), and reaction type (upvote/downvote, like, dislike, none, stars). 3. Rules: mandatory timed reading of 6 rules (min 15 seconds). 4. Review: text preview, settings summary, confirmation, and submit for approval.

14 types in 5 groups: Lyrics: published text, work in progress, sample Collaboration: collaboration (looking for / offering) Feedback: review Discussion: discussion, question, tips & tricks, inspiration, tutorial, challenge Other: announcement, poll, off topic Categories change dynamically based on type — lyrics use genres, discussions use topics (technique, theory, production, history, trends, experiences, guides, other).

When creating a post, you can switch to anonymous mode. Your post will appear under the name "Anonym" with a "?" avatar. You can set a 4-digit PIN for later identity verification when requesting deletion via support. Anonymous post restrictions: • Comments and critiques are automatically disabled • Copyright type is forced to Copyright Free • A one-way IP hash is stored (max 10 posts per day)

Every post must be approved by an admin or moderator before appearing in the feed. Content, media links, and plagiarism are checked during approval. Post statuses: • Pending — just submitted, visible only to author with "Pending approval" message • Approved — visible to everyone • Rejected — denied by admin • Deleted — soft-deleted by admin or author

The post author chooses the reaction type when creating: • Upvote / Downvote — ▲ / ▼, net score displayed • Like with reason — 👍 with reason selection • Like only — just 👍 • Dislike only — just 👎 • None — reactions hidden • Stars 1-5 — star rating The feed always shows a universal heart toggle (♡/♥). The detail page shows reactions based on the chosen type.

Comments and critiques are controlled by a single toggle (author can enable or disable them). • Critiques — longer feedback (min 20 chars), can be deleted by post author or critique author • Comments — free comments with no minimum length Each comment has 3 emoji buttons: 👍 like, 👎 dislike, ❤ heart. Clicking the same button removes the reaction, clicking another switches it. Login is required for commenting and reacting.

On the post detail page there is a "Report" button. You can report posts, comments, or critiques. Enter a reason and the report goes to the admin panel for moderators to resolve.

Yes. The right sidebar has a "My Profile" section with "Manage Posts" button. Opens a dashboard with all your posts. For each post you can: • Edit — change title and text (inline editor) • View — go to detail • Delete — two-step confirmation In profile settings you can change your display name (applies to existing posts too).

The left sidebar has a search field. You can toggle between: • Text — searches approved post content • User — searches by display name Search is debounced (300ms), minimum 2 characters, returns max 20 results.

Before your first post, you must agree to 6 rules (with a mandatory timer): 1. Respect others 2. Write your own lyrics 3. No spam 4. No AI-generated content 5. No hate speech 6. Respect copyright Consent is stored in localStorage. Violations may result in post rejection or deletion by moderators.

Media links allow adding URLs to images or music platforms to your post. Supported platforms: YouTube, Spotify, SoundCloud, Suno, Udio. Images must be direct URLs (https, .jpg/.png/.gif/.webp extension), max 2048 characters. All links start as unapproved (pending admin review). When a post is approved, its links are auto-approved. Only approved links are shown on the post detail page with platform-specific icons.

The right sidebar shows: • Overall stats — total posts, users, and comments • Top contributors — 5 most active users • Genres — 8 most used genres by frequency The main feed can be sorted by: latest, most popular (by like count), or most viewed (by views count).

A professional book writing tool for browser with chapter tree structure, character database, world building, timeline, and PDF, DOCX, EPUB export.

Click the Create Project button in the sidebar. Enter the project name and author name. The project is automatically created with a first chapter.

Yes. Use the Export Manager to export your book. We support PDF (A4), DOCX, and EPUB (e-books).

Book Creator supports organizing chapters in a tree hierarchy. Each chapter can have unlimited sub-chapters, making it easy to structure complex stories.

In the Characters section you can create character profiles with attributes like name, age, description, motivation, and relationships. The database is linked to chapters for easy reference.

Yes, the basic version of Book Creator is free with unlimited projects.

Book Creator requires an internet connection for saving and loading projects. Writing and editing happen locally in your browser.

Yes, The Forge is a Progressive Web App (PWA) — you can install it to your home screen and use it in standalone mode. For serious writing and the Pitch Engine, we recommend a desktop computer (due to large dictionaries, microphone requirements, and full keyboard needs).

Modern browsers with Web Audio API: Chrome, Firefox, Edge, Safari. Requires getUserMedia (microphone), IndexedDB, Web Crypto API and localStorage. IE is not supported.

Yes. The Forge is a fully functional Progressive Web App. You can install it to your home screen (via the browser's "Add to Home Screen" option) and use it as a standalone app without the browser address bar. It includes a service worker for offline caching and faster loading.

The Forge has a layered performance system that auto‑adapts to your device: Profiles (Potato → School → Handle → NASA): Each profile controls 16 settings across 5 categories (Graphics, Physics, Audio, Animation, Interface). Higher profiles = more visual effects and more accurate computations. Benchmark: Tests 8 real app algorithms (YIN detection, physics, canvas, Levenshtein rhymes, FFT beat, DOM, layout) and produces a 0–100 score. The result auto‑selects the optimal profile. WebAssembly (Wasm) with SIMD128: YIN pitch detection is compiled from C++ to Wasm. 2–4× faster than JavaScript, module is only 8 KB. Lazy‑loaded on first Pitch Engine use. Worker Pool: Background computations in dedicated threads. Worker count = CPU cores − 1 (NASA), 0 (Potato). Three modes: persistent (YIN detection, worker stays alive), run‑or‑drop (skip delayed frames), one‑shot (rhyme analysis, worker terminates with result). Fallback chain: Wasm → Worker pool → JS inline. If a layer fails, the next one takes over.

Main reasons: Phonetic dictionaries (~4 MB): On first rhyme lookup, the 9.7 MB Czech trie dictionary is loaded — this causes a brief freeze. Pitch Engine (~100ms latency): Audio processing and YIN detection have natural latency. The Wasm SIMD version is 2–4× faster than the original JS. CSS bundle (~10,000+ lines): The custom CSS engine is large, optimization is planned. Recommendation: if the app is slow, try switching Performance mode in profile settings to a lower level (Potato, School, Handle, NASA). Lower profiles disable expensive effects and reduce worker count. Newer devices and browsers generally provide better performance.

The rhyme engine, stress analysis, and dictionary lookup run server-side. This is to protect intellectual property — these algorithms are unique and their source code is a trade secret. The server API does not collect any data — it only performs calculations and returns results. All your lyrics, projects and recordings stay locally in your browser.

Most features are fully polished and ready to use. Here's the real status: Complete (100%): Block Editor (all modes), Notebook Mode, Focus Mode, auto-versioning, Rhyme Engine (CZ + EN), scheme and technique detection, Knowledge Book, Inspiration Templates, Today's Spark, Word Bank, Reference Panel, Mood Matcher, all 48+ genre templates, Export (8 formats, 12 themes, 16 fonts), Metronome, Audio Player, Story Archive, Profile (16 sections), Level system and achievements, Bin (Trash), PWA, language switch, App Tour, Cookie Consent, decorative backgrounds, animations, performance profiles, Web Worker pool, Wasm SIMD128 YIN detection. In progress (~80–95%): Pitch Engine — detection and live metrics work, some training modes are being refined. Creators Wall — core feed, posts and comments are done, moderation and search are being iterated on. Cloud sync (Dropbox/Drive) — functional but occasionally needs manual reconnection. Help section — covers basics, being expanded gradually. Does everything run smoothly? Yes. The Forge runs mostly client-side — data is in localStorage and IndexedDB, audio is processed by the Web Audio API directly in the browser. The only server calls are lightweight API requests for rhyming and dictionary lookups — they don't slow the app down. After the first load (when dictionaries and CSS are downloaded), everything is fast and responsive. The app is built as a single-page application — navigation between pages is instant, no full page reloads. The only potential freeze is the initial rhyme dictionary load (~9.7 MB), but that's a one-time thing and it stays cached afterwards.

No. The Forge is deliberately purely rule-based. All features are built on deterministic algorithms: G2P phonetics, YIN pitch detection, rhyme scoring, voice classification, technique detection. No LLMs, no machine learning models, no artificial intelligence. This is a conscious design choice.

The app uses API endpoints for rhyme detection, dictionary lookup, cloud sync and originality checking. APIs run server-side (not in the browser) and are protected against abuse.

Yes. All your data stays locally in your browser — it never leaves your device unless you choose to use cloud sync. The app has no server-side database that stores your lyrics or recordings. API communication is encrypted.

The profile and export can be encrypted (AES-256-GCM) with strong key derivation from a password. Emergency backup (.forge file) is a self-contained encrypted archive with all data.

Multi-layer protection at both network and application level: Network layer: DDoS protection, WAF, SSL, CAPTCHA. Server layer: Security headers (HSTS, CSP), CORS restrictions. Application layer: Rate limiting, obfuscated code, input sanitization, anti-bot protection.

Yes. The robots.txt blocks GPTBot, ChatGPT-User, CCBot, Anthropic, Google-Extended and all major AI crawler bots.

Essential: Vercel (hosting), Cloudflare (CDN + security). Optional: Dropbox API, Google Drive API, PostHog (anonymous analytics — activated only after consent). Feature: DuckDuckGo API, Free Dictionary API, Google Fonts.

Write to theforgewrite@gmail.com. The app does not yet have a built-in feedback form or public issue tracker. You can also use the same email for licensing inquiries and collaboration requests.

The Forge is not open source and is not open to external contributions. The repository is private. If you'd like to support development financially, you can contribute via Buy Me a Coffee or PayPal — links are in the page footer.

Development follows the roadmap on the home page. The main planned phases: 1. Audio Workspace & Enhanced Writing: Voice memos linked to lyrics, beat/instrumental import, version history with draft comparison, project hub for albums and collections, built-in dictionary and thesaurus, rhyme analytics dashboard, PDF/DOCX/LRC export, zen mode, PWA mobile version with offline mode (limited but usable). 2. Poetry Mode: Meter and verse rhythm analysis, poetic syllable counting and mapping, literary figure detection (imagery, metaphors, alliteration, anaphora), templates for sonnet, haiku, villanelle and free verse. 3. Book Mode: Project Tree with hierarchical structure, Research & Storyline with fact database, event timeline, Whole Project View, Mini DTP editor with typesetting and covers, PDF export (A4, A5, Letter, Booklet). 4. Screenplay Mode: Auto formatting per industry standards, scene timeline and continuity overview, character database and dialogue statistics, film PDF export. Final state — The Creative Forge: Seven connected modes (Beginner, Block Editor, Notebook, Poetry, Book, Pitch, Screenplay) covering the entire process from first idea to final work.

The version is shown in the page footer and in JSON-LD structured data. The development roadmap is available on the home page.