The professional media toolkit.
Download from 1000+ sites. Transcribe with local Whisper AI. Convert formats with GPU acceleration. Clip, compress, batch process. Everything runs on your machine.
# Paste a URL. Siphon handles the rest. [Siphon] Fetching media info... [Siphon] Source: youtube.com [Siphon] Format: 1080p MP4 | Duration: 12:34 [Siphon] Downloading... 87% | 42.1 MB/s | ETA 3s [Siphon] Complete → ~/Downloads/video.mp4 # One-click transcription with local Whisper [Siphon] Transcribing with whisper-large-v3 [Siphon] 1,847 words → video.srt
Every media operation you need, in one place. No CLI chains, no browser extensions, no subscriptions.
YouTube, TikTok, Instagram, Twitter, and 1000+ sites. Parallel chunked downloads saturate your connection.
Local OpenAI Whisper integration. Generate SRT/VTT subtitles from any media. No cloud, no subscription.
H.265, AV1, VP9, ProRes, and more. Hardware-accelerated encoding via NVENC, QSV, or AMF.
Frame-accurate clip extraction by timestamps. Lossless cuts when possible, re-encode when necessary.
Reduce file size without visible quality loss. Configurable quality presets from minimal to aggressive.
Convert video segments to high-quality GIFs. Custom FPS, resolution, and duration controls.
Pull audio tracks from any video. Export as MP3, AAC, FLAC, WAV, or OGG.
Extract individual frames at custom intervals. PNG or JPEG output for thumbnails and stills.
Queue multiple files for any operation. Convert, compress, or extract in bulk with one click.
A Rust core that starts in milliseconds and processes media at hardware limits.
No Electron. No Python runtime. The entire backend is compiled Rust (Tauri 2.0), producing a lightweight binary that uses a fraction of the memory.
Auto-detects NVIDIA NVENC, Intel QSV, and AMD AMF hardware encoders. Falls back gracefully to CPU when unavailable.
Multi-threaded chunked fetcher saturates your connection. Configurable parallel threads with queue management.
A production-grade UI built with React 19, Framer Motion, and a premium design system.
Paste URLs, pick formats, monitor progress. Real-time speed, ETA, and status for every job.
Convert, clip, compress, extract — all from a unified panel. Drag files in, pick options, click go.
Select a Whisper model, point at a file, generate subtitles. Language detection, word timestamps.
Output directories, hardware encoder selection, quality presets, thread count. Everything configurable.
No cloud. No telemetry. No accounts. Every operation runs on your hardware.
Downloads, transcriptions, conversions — everything processed on your machine. Nothing is uploaded anywhere.
No sign-up, no login, no cloud sync. Download, install, use. That's it.
Full source code available. Audit every line, fork it, contribute back.
Works without internet for all local operations. No license servers, no remote dependencies.
Download, transcribe, convert, clip, compress — all in one app that respects your machine and your privacy.