English | 繁體中文

PodNote

Turn Podcast / YouTube content into transcripts, then let Gemini organize them into structured notes.

Two Editions (Web / Desktop)

PodNote comes in a Web edition and a Desktop edition, with different capabilities and API key requirements:

	Web	Desktop
Platform	Web + Android	Windows desktop
Audio sources	Podcast	YouTube / Podcast / uploaded audio files
Transcription	Backend faster-whisper (runs on the server) or external Groq (faster)	Cloud Whisper API (Groq / OpenAI)
Notes	Gemini	Gemini
API keys	Gemini (required) Groq (optional, for fast transcription)	Groq or OpenAI (required) + Gemini (required)
Installation	None — open the live app → https://frankkn-podnote-api.hf.space	Download the installer.exe (Windows)

Desktop Screenshots

Open the Web App Now

The fastest way to get started: no installation, just open the deployed live version.

https://frankkn-podnote-api.hf.space

The first visit may take a few seconds while the Hugging Face Space wakes up. Once the app is open, go to the Settings page and enter your own Gemini API Key (used for generating notes); if you want "Fast" transcription, add your own Groq API Key as well. After that, just paste a Podcast link to generate notes.

There are two transcription modes:

• Fast (recommended): Transcribes on an external GPU via the Groq Whisper API — quick and accurate. You need to enter your own Groq API Key and Gemini API Key on the Settings page (see the API key guide below — it's really easy!). Keys are stored locally on your device; they are only sent to the backend to call Groq when you generate notes, used once, and never stored.
• Slow (simple): Transcribes on the server's CPU. Only a Gemini API Key is required, but it is slower and subject to length and concurrency limits.

Service health check:

https://frankkn-podnote-api.hf.space/health

Use the Desktop App Now

The desktop edition is a standalone Windows application — no backend to set up, just download and install. It bundles yt-dlp and fetches ffmpeg automatically, so it can handle YouTube / Podcast links or local audio files directly. Transcription goes through cloud Whisper (Groq or OpenAI), and notes are generated with Gemini.

Download and Install

1. Open the GitHub Releases page.
2. Under the latest release (tagged Latest), download the installer PodNote-Setup-X.Y.Z.exe from the Assets section.
3. Double-click the installer. If Windows SmartScreen shows "Windows protected your PC", click "More info" → "Run anyway" (a normal prompt for unsigned applications).
4. After installation, launch PodNote from the Start Menu or the desktop shortcut.

First-Time Setup

Once the app is open, go to the Settings page and enter your API keys (see "Getting API Keys" below for how to obtain them):

• Gemini API Key (for generating notes, required)
• Groq or OpenAI API Key (for cloud Whisper transcription, required; either one works — if both are set, OpenAI takes priority)

Keys are stored locally on your device; they are only attached to requests when you generate notes, used once, and never stored.

Getting Started

Back on the main screen, paste a YouTube / Podcast link or pick a local audio file, then hit Start. The app automatically downloads the audio, transcribes it, and organizes the transcript into structured notes with Gemini.

Auto-update: The desktop edition ships with electron-updater. On startup it checks GitHub Releases for a new version, downloads it in the background if available, and prompts you to update.

Getting API Keys

Gemini API Key (required)

Used to turn transcripts into notes. The free tier is usually more than enough for personal use.

1. Open https://aistudio.google.com and sign in with your Google account.
2. Click "Get API key" in the left sidebar.
3. Click "Create API key" → pick a Google Cloud project (or let it create one automatically) → click "Create API key".
4. Copy the key shown on screen (it starts with AIza).
5. In the app's Settings page, paste the key into the "Gemini API Key" field and save.

If you hit a 429 quota error, the key's free quota for a specific model has usually run out. You can go to aistudio.google.com and create a new key under a new Google project.

---

Groq API Key (for Fast mode, optional)

Only needed if you choose the "Fast (recommended)" transcription mode. Currently free, no credit card required.

1. Open https://console.groq.com and sign in / register with a Google account or email.
2. Click "API Keys" in the top-right menu.
3. Click "Create API Key", give it a name (e.g. podnote), and create it.
4. Copy the key shown on screen (it starts with gsk_). The key is shown only once — copy it immediately.
5. In the app's Settings page, paste the key into the "Groq API Key" field and save.

No Groq key? No problem — switch to the "Slow (simple)" mode and you're good to go, no extra setup needed.

---

Project Structure

PodNote/
  backend/   FastAPI API, transcription jobs, exported web static files (web edition deployed to HF Space)
  frontend/  Expo / React Native Web frontend source (web edition UI)
  desktop/   Electron desktop app source (Windows, packaged standalone, no backend dependency)

Changelog

Web

Version	Date	Highlights
v1.4.0	2026-06-09	Added a user guide page; backend now supports OpenAI Whisper transcription (as an alternative to Groq); fixed downloaded audio not being cleaned up after transcription, which caused server disk usage to grow episode by episode
v1.3.1	2026-06-01	Settings page shows the current version; version info centralized (app.json as the single source of truth); re-exported and redeployed static files
v1.3.0	2026-06-01	Auto-generate a summary with Gemini in the background after transcription; Chinese output fixed to Traditional (OpenCC s2tw); summary flow moved to module level for stability
v1.2.0	2026-06-01	Fixed blank screen in the web desktop layout (emoji instead of vector-icons, Sidebar switched to router.push, cards stretched with flex); added Playwright tests
v1.1.0	2026-06-01	Two-column Sidebar layout for wide screens (≥768px); Markdown rendering for notes; long episodes split into 5-minute segments for transcoding; abuse prevention and audio length limits
v1.0.0	2026-06-01	First stable release: Podcast link → transcript → Gemini notes; dual-mode transcription (Groq fast / CPU slow); concurrent episodes and note history; deployed on Hugging Face Spaces

Desktop

Version	Date	Highlights
v1.3.0	2026-06-09	Remote transcription now uploads in segments (files over 24MB are split by ffmpeg into 16kHz mono chunks), so long episodes no longer fail on the API's single-file limit; prompts to install ffmpeg and retries automatically if it is missing
v1.2.0	2026-06-09	Switched to GitHub Actions (Windows runner) for automated packaging and releases; web / desktop releases split by tag prefix so they don't interfere; installer name fixed to PodNote-Setup-<version>.exe
v0.2.0	2026-06-04	Download button placed on the same row as the URL input; transcript box reduced to 2 lines; Settings page marks the transcription API key as required
v0.1.1	2026-06-04	First standalone Windows desktop release: no backend needed, packaged as an installer; bundled yt-dlp with automatic ffmpeg download on first run; three-column Sidebar layout; Gemini note generation; auto-update via GitHub (electron-updater)