Freedom251 › AI Video Caption Generators

VidScribe

Free plan Self-hostedWeb

What VidScribe is

VidScribe is a free, open-source application for transcribing videos, YouTube URLs, and audio locally. It uses Whisper for automatic language detection across 99 languages and accepts FFmpeg-supported formats such as MP4, MKV, AVI, MP3, and WAV. Transcripts export as SRT, VTT, TXT, or JSON. Processing runs on a CPU or GPU and can work offline after setup and model download. The project says no cloud services or API keys are used and that data stays on the computer. Users can select tiny, base, small, medium, or large Whisper models to balance speed and accuracy. NVIDIA CUDA, Apple Silicon MPS, and CPU-only processing are supported. VidScribe is self-hosted with a web interface and documents REST endpoints for transcription and exports plus a WebSocket progress endpoint. Setup requires Node.js 20+, Python 3.11+, pnpm, and FFmpeg.

Who it suits

It suits users who need local, offline-capable transcription and developers wanting a self-hosted web interface or documented API. It also fits workflows requiring multiple subtitle and text export formats.

What it does well

Supports transcription in 99 languages
Keeps processing on the local computer
Works with CPU, CUDA, or Apple Silicon MPS
Exports SRT, VTT, TXT, and JSON
Offers multiple Whisper model sizes

What to know first

Requires Node.js, Python, pnpm, and FFmpeg
Large model may need about 10 GB GPU memory
Setup and model download are required for offline use

Plans

Plan	Billed and limits	Price
Free	Free forever · no cloud services or API keys stated	Free

As github.com lists them, read 12 Oct 2026. The maker's own page has the final word.

Verdict

VidScribe focuses on private, local transcription with broad language and format support. Its self-hosted setup and hardware requirements are the main considerations before use.

Facts from VidScribe's own pages

API: The project documents REST endpoints for starting and retrieving transcriptions and exporting results, plus a WebSocket progress endpoint. · github.com, 12 Oct 2026
Caption languages: 99 · github.com, 22 Sept 2026
Exports: Transcriptions can be exported as SRT, VTT, TXT, or JSON. · github.com, 12 Oct 2026
Free plan: Yes · github.com, 22 Sept 2026
GPU support: It supports NVIDIA CUDA and Apple Silicon MPS acceleration, as well as CPU-only processing. · github.com, 12 Oct 2026
Integrations: The README names OpenAI Whisper for transcription and yt-dlp for downloading audio from YouTube URLs. · github.com, 12 Oct 2026
Languages: It supports 99 languages with automatic language detection using Whisper. · github.com, 12 Oct 2026
Maker: The GitHub profile identifies Isaque Coelho and lists Brazil as the profile location. · github.com, 12 Oct 2026
Media input: It accepts YouTube URLs and audio or video formats supported by FFmpeg, including MP4, MKV, AVI, MP3, and WAV. · github.com, 12 Oct 2026
Model choices: Users can choose Whisper tiny, base, small, medium, or large models to balance speed and accuracy. · github.com, 12 Oct 2026
Notable limits: The README lists approximate GPU memory requirements from about 1 GB for tiny and base models to about 10 GB for the large model. · github.com, 12 Oct 2026
Privacy: The project says it uses no cloud services or API keys and that no data leaves the user's computer. · github.com, 12 Oct 2026
Processing: The README says transcription runs locally on the user's CPU or GPU and works offline after setup and model download. · github.com, 12 Oct 2026
Purpose: VidScribe is an open-source application for transcribing videos, YouTube URLs, and audio locally. · github.com, 12 Oct 2026
Requirements: The listed prerequisites are Node.js 20+, Python 3.11+, pnpm, and FFmpeg. · github.com, 12 Oct 2026
Subtitle export: other formats · github.com, 22 Sept 2026
Support: The README invites users to open an issue or submit a pull request to contribute. · github.com, 12 Oct 2026

Where VidScribe stands on Freedom251

AI Video Caption Generators#74 of 99