Stop renting software.

← All tools · AI transcription

Open source alternatives to Otter.ai

Otter.ai starts at $8 (pricing). Here are 3 self-hostable options – free to run on your own server.

whisper

★ 110.5kMIT
self-host: medium

openai/whisper

faster-whisper

★ 25.9kMIT
self-host: medium

SYSTRAN/faster-whisper

whisper.cpp

★ 54.4kMIT
self-host: medium

ggml-org/whisper.cpp

Build it yourself with AI

Feasible: Whisper + LLM summary pipeline is a classic AI build.

Paste this prompt into Cursor, Claude Code, Codex or any coding agent to build a minimal self-hosted Otter.ai clone, deployable with Docker on a VPS.

You are a senior full-stack engineer. Build a minimal, self-hosted clone of Otter.ai (AI transcription) that I can run on my own VPS. Keep it small, production-sane, and easy to maintain.

## Goal
Replace my paid Otter.ai subscription for a single person or small team. MVP only — no billing, no enterprise features.

## Core features (build exactly these)
1. audio/video upload and browser recording
2. transcription with faster-whisper
3. speaker diarization (pyannote)
4. LLM summary and action items (open-weight model via Ollama)
5. searchable transcripts with timestamps

## Stack
- Next.js (App Router, TypeScript) + Tailwind + shadcn/ui
- PostgreSQL with Drizzle ORM (migrations checked in)
- Auth: Better Auth (email + password, optional GitHub OAuth); every query scoped to the signed-in user/workspace
- Background jobs (if needed): pg-boss on the same Postgres
- File storage (if needed): S3-compatible (MinIO container)

## Data model (start here, extend only if required)
User, Recording, Transcript, Segment, Summary
Use UUID primary keys, createdAt/updatedAt on every table, and foreign keys with sensible ON DELETE behavior.

## Deliverables
1. Working app with the features above and a clean, minimal UI
2. docker-compose.yml with app, postgres, minio and Caddy (automatic HTTPS) — `docker compose up -d` must be all I run
3. .env.example documenting every variable; no secrets in code
4. Seed script with demo data
5. README: local dev, deploy on a fresh Ubuntu VPS (install Docker, set DNS A record, compose up), backups (pg_dump cron), and updating

## Rules
- Build in small steps; after each feature, run typecheck and the app, and fix errors before moving on
- Validate all input with zod; rate-limit public endpoints
- Write a few Playwright smoke tests for the main flow
- Ask me before adding any dependency beyond the stack above

For reference (study the approach, don't copy code): https://github.com/openai/whisper, https://github.com/SYSTRAN/faster-whisper, https://github.com/ggml-org/whisper.cpp

One self-hosted swap per week

A new SaaS teardown and its open-source replacement, every week. No spam.

We only use your email to send the SelfHostThis newsletter (stored with our email provider Resend, EU). Every email has a one-click unsubscribe link, and you can ask us to delete your data anytime. We use double opt-in. Datenschutz · Privacy policy