I built a free, local video transcription tool, because I didn't want to pay $10/hour or upload my files to a stranger's server

Giuseppe Carlà — Sat, 09 May 2026 16:10:14 +0000

Every time I needed to transcribe a video at work, I hit the same wall:
the good tools cost money per minute, and the free ones upload your files
to a remote server. Neither was acceptable for work content.

So I built "Pitchfall" - a local transcription tool that runs entirely
on your own machine.

What it does

Upload any video or audio file (or paste a YouTube URL), and Pitchfall:

Transcribes it locally using faster-whisper
Shows a real-time progress bar with the current segment being recognized
Syncs the transcript to the video — click any line to jump to that moment
Exports as .txt or .srt subtitle file
Optionally translates into 10 languages via OpenRouter free models

No API key needed for transcription. No account. No cloud.

The stack

faster-whisper (local Whisper model)
│
▼ streaming SSE
FastAPI (Python)
│
▼
Next.js 16 + Tailwind CSS 4

The backend streams transcription progress via Server-Sent Events —
each segment gets sent to the frontend as it's recognized, so you see
the text appear in real time rather than waiting for the whole file
to finish.

Why local matters more than I expected

When I started this I thought "local vs cloud" was mainly a cost issue.
It turned out to be a correctness issue too.

Faster-whisper on CPU with the small model is genuinely fast enough
for practical use — a 5-minute video takes about 2-3 minutes on a
mid-range laptop. More importantly, the transcript never touches a
third-party server. For work content, legal recordings, or anything
sensitive, that distinction matters.

The part that took longest: memory management

The original version leaked memory on every transcription. The culprit
was URL.createObjectURL() — a blob URL that keeps the entire video
file in RAM. It was never revoked, so after 3-4 sessions the browser
was holding multiple full videos in memory.

The fix is a single line, but finding it required profiling:

// Before reset, always revoke the previous blob URL
if (isBlobUrl && mediaUrl) URL.revokeObjectURL(mediaUrl);

The backend had a similar problem: temp files from crashed SSE
connections weren't getting cleaned up. I solved it with a FastAPI
lifespan context manager that wipes .tmp/ on startup and shutdown.

What I'm less happy with

Translation reliability. The free OpenRouter models have rate limits
and occasionally go offline. Pitchfall tries 5 models in order with
automatic fallback, but if they're all saturated you get a 503. For
casual use it's fine; for production you'd want a paid model.

No GPU support in the Docker image. The Dockerfile uses CPU-only
inference. Adding CUDA support means a much heavier image and
nvidia-container-toolkit as a prerequisite — I left it out for now
to keep the setup simple.

YouTube sync limitation. For uploaded files, clicking a transcript
segment seeks the video instantly. For YouTube URLs, the video loads
as an iframe embed — the YouTube API doesn't allow external seek
control without a more complex integration.

Try it

GitHub: https://github.com/scibilo/pitchfall

Manual setup takes about 5 minutes if you have Python 3.10+ and Node.js
18+ installed. Docker setup is one command.

The only hard dependency that people often don't have: ffmpeg.
sudo apt install ffmpeg on Ubuntu, brew install ffmpeg on Mac.

I'm curious: do you handle transcription in any of your projects?
What's your current setup — local model, cloud API, or something else?

Telepage – I built a self-hosted PHP app that turns any Telegram channel into a website

Giuseppe Carlà — Wed, 01 Apr 2026 17:55:48 +0000

If you run a Telegram channel, you already know the problem: your content is invisible to Google, there's no search, old posts are buried, and readers need the app just to see your work.

I built Telepage to fix that.

What it does

Telepage connects to your Telegram channel via a bot webhook and turns every post into a searchable web card — automatically, in real time.

Every hashtag in your Telegram posts becomes a colored navigation filter. Every link gets its Open Graph metadata scraped. Every post gets an AI-generated summary and tags if you connect a Gemini key.

The tech stack

Pure PHP 8.1, SQLite with WAL mode, vanilla JS. No frameworks, no Composer, no build step, no MySQL. It runs on standard shared hosting — I tested it on Aruba (a very restrictive Italian host).

Telegram channel
      │
      ▼ webhook (instant)
PHP 8.1 + SQLite
      │
      ▼
Your website — card grid, search, tag filters

Interesting technical decisions

Session isolation per installation
Multiple Telepage sites on the same domain (e.g. site.com/news/ and site.com/recipes/) need completely separate admin sessions. I solved this with:

session_name('tp_' . substr(md5(TELEPAGE_ROOT), 0, 12));
session_start();

Each installation path produces a unique session name — no shared cookies, no cross-login.

History Scanner
Telegram's Bot API has no "get all past messages" endpoint. To import historical content I use the forwardMessage trick: forward each message ID from the channel to itself, read the content, then immediately delete the forwarded copy. It scans backwards from the most recent ID, skipping gaps from deleted messages.

AI integration
Optional Google Gemini integration auto-tags and summarizes every post. The models available via the free tier change frequently — I built a cascade fallback that tries multiple model names in order and logs exactly which one succeeded.

What it looks like in production

I've been running it on two test channels:

A science/news channel: 23 posts, tagged by topic
A recipes channel: 952 posts, fully tagged and summarized by AI

The recipes site went from zero to 952 searchable, tagged posts in a few hours using the History Scanner.

What I'm less happy with

AI calls are currently synchronous in the admin panel — for large archives you click "Process AI" repeatedly. A proper background queue would be better.
The History Scanner requires manual ID tuning when posts are missing — not ideal for non-technical users.
No pagination on the install wizard, though the 5-step flow works fine in practice.

Try it

GitHub: github.com/scibilo/telepage

It's MIT licensed. Works on any PHP 8.1+ shared hosting with HTTPS. The install wizard takes about 5 minutes.

Feedback welcome — this is the first public release and I'm actively improving it.

Forem: Giuseppe Carlà

I built a free, local video transcription tool, because I didn't want to pay $10/hour or upload my files to a stranger's server

What it does

The stack

Why local matters more than I expected

The part that took longest: memory management

What I'm less happy with

Try it

Telepage – I built a self-hosted PHP app that turns any Telegram channel into a website

What it does

The tech stack

Interesting technical decisions

What it looks like in production

What I'm less happy with

Try it