ActorRise
A platform for actors, built with Next.js and FastAPI. Find audition material in
plain English, then actually rehearse it.
Live at actorrise.com.
What's in the library
| monologues | titles |
|---|
| Plays | 13,998 | 732 |
| Film | 4,069 | 1,208 |
| Television | 1,313 | 224 |
| Searchable total | 19,380 | |
Every piece is 100+ words for stage and 75+ for screen (see
monologue_quality.min_words_for_source — the floor is per-source and lives in
one place). Full text is stored only where the rights allow it; everything else
links out to where an actor can buy the script.
A known gap, stated plainly: the play corpus is public domain, which means
nothing written after 1929. There is no contemporary play in the library, and no
amount of scraping changes that, because every contemporary play is in
copyright. 8% of searches ask for contemporary work.
Features
- Monologue search — semantic search over the whole library. Understands
duration ("under 90 seconds"), act and scene, character names, tone, age, and
gender. Filters for overdone pieces so you are not the fourth Hamlet that day.
- ScenePartner — AI scene reader that runs the other parts so you can
rehearse a scene alone. Upload a script, it extracts the scenes.
- Monologue Work — audio-first rehearsal for a single piece, with
line-by-line delivery tracking.
- Callboard / Green Room — a shared board and activity feed.
- Auditions, resume, self-tapes — track submissions, build a resume, store
tapes.
- Actor profiles — headshot, age range, type and training, used to bias
search results toward what suits you.
Tech stack
Frontend — Next.js 16 (App Router), React 19, TypeScript, Tailwind CSS v4,
shadcn/ui, Framer Motion, Zod 4. Deployed on Vercel.
Backend — FastAPI, SQLAlchemy, Pydantic, uv. Deployed on Render
(auto-deploys on pushes touching backend/**).
Data — Supabase Postgres with pgvector. Embeddings are
text-embedding-3-large at 1536 dimensions, searched with cosine distance over
an HNSW index. gpt-4o-mini handles query parsing and content analysis.
Auth and storage — Supabase.
Getting started
Prerequisites: Node 18+, Python 3.9+, uv, a
Postgres database with the vector extension, and a Supabase project.
git clone <your-repo-url> && cd actorrise
# Frontend
npm install
npm run dev # http://localhost:3000
# Backend, in a second terminal
cd backend
uv pip install -e .
uv run uvicorn app.main:app --reload # http://localhost:8000
Create .env.local and backend/.env by hand from the blocks below. There are
no .env.example files to copy: .gitignore excludes .env*, so a committed
example would be invisible to everyone who cloned the repo.
Tables are created by SQLAlchemy on first startup.
Environment
Frontend, .env.local
NEXT_PUBLIC_API_URL=http://localhost:8000
NEXT_PUBLIC_SITE_URL=http://localhost:3000
NEXT_PUBLIC_SUPABASE_URL=
NEXT_PUBLIC_SUPABASE_ANON_KEY=
# NEXT_PUBLIC_GA_MEASUREMENT_ID=G-XXXXXXXXXX # optional
Backend, .env (the settings app/core/config.py reads)
DATABASE_URL=postgresql://user:password@localhost:5432/actorrise
JWT_SECRET=
JWT_ALGORITHM=HS256
CORS_ORIGINS=http://localhost:3000
SUPABASE_URL=
SUPABASE_SERVICE_ROLE_KEY=
SUPABASE_STORAGE_BUCKET=headshots
OPENAI_API_KEY=
ENVIRONMENT=development
SUPERUSER_EMAILS=
REVIEW_EMAILS=
Layout
app/ Next.js App Router
(auth)/ login, signup
(marketing)/ public pages, SEO collection pages
(platform)/ monologues, practice, rehearse, scenes, callboard,
greenroom, audition, resume, profile, billing, admin
components/ React components (search/, profile/, ui/, ...)
lib/ api client, auth context, Supabase client
backend/
app/api/ route modules, incl. admin/
app/models/ SQLAlchemy models
app/services/
search/ semantic_search, title_lookup, query_optimizer
extraction/ TEI XML, Gutenberg plain text, screenplay PDF parsers
data_ingestion/ pipeline.py — the only supported way to add monologues
ai/ embeddings, content analysis
scripts/ corpus maintenance and audits (see below)
tests/ 119 files, 1,537 tests
Working on the corpus
ingest_play() in app/services/data_ingestion/pipeline.py is the only
supported way to add monologues. It enforces the rights model
(services/licensing.py), the word floor, and cross-source deduplication.
Useful scripts:
python -m scripts.audit_corpus_quality # run after EVERY ingest
python -m scripts.audit_search_constraints # replay real searches, check results
python -m scripts.apply_table_storage_params # DB storage params, in version control
Two habits worth keeping:
- Run the quality audit after every ingest. It found a Plautus footnote bug
in minutes on a source nobody had read, where every other problem cost hours
of sampling and luck.
- Do not scan the whole corpus from a laptop. Supabase bills egress and the
maintenance scripts are the largest consumer. Select the columns you need and
page; push work into SQL where you can.
Testing
cd backend && pytest -q # 1,537 tests
npm run lint # frontend
License
MIT
