Every word you speak, instantly captured.
Record a demo, walkthrough, or tutorial. The moment you stop, Hosaka starts transcribing - locally, automatically, in the background. Your video and captions are ready together.
Record. Hosaka does the rest.
Stop recording. Your transcript starts automatically. By the time you've trimmed your video, it's done.
Record
Hosaka captures your screen, webcam, and mic together. Just hit record.
Transcribe
When recording stops, transcription starts automatically in the background. faster-whisper runs locally on your CPU - no GPU required.
Export
Your video, captions file, and full transcript - ready together. No extra steps.
Your words stay on your machine.
Transcription runs locally using Whisper. We built it that way because we think your words belong to you - not to a cloud service, not to a metered API.
No cloud.
Transcription runs locally using Whisper via CTranslate2. Your audio is never uploaded anywhere.
No account.
No API keys, no usage quotas, no service that can go down or change its pricing.
No limits.
Transcribe every recording you ever make. No monthly minute caps, no subscription tiers. The model downloads once; after that it's yours forever.
What you get with every recording.
Automatic
Starts the moment recording stops. There's nothing to click, nothing to configure. It just happens.
SRT with every export
The subtitle file ships alongside your video - ready to upload to YouTube, embed in a player, or import into any editor.
Works offline
After the first model download (~150 MB), transcription works with no internet connection, every time.
Non-blocking
Transcription runs in the background while you edit. Export waits until it's done, then everything lands together - video, captions, transcript.
Recordings that work harder for you.
Transcription isn't just a file - it's what turns a recording into something searchable, accessible, and ready to publish.
Captions on YouTube, instantly
Upload your video and SRT file - YouTube auto-syncs them. No extra tools, no manual timing.
Your words become searchable
Every recording you make is a transcript. Search your own library the way you search documents.
Accessible by default
Captions aren't an afterthought. Every recording you export is ready for viewers who need them.
Software you actually own.
One payment. No subscription cloud account required. Hosaka runs entirely on your hardware.
30-day money-back guarantee
30-day money-back guarantee
Questions
What languages are supported?
Whisper supports over 90 languages. Hosaka auto-detects the language spoken - no configuration needed.
How accurate is it?
Accuracy depends on microphone quality, background noise, and how clearly you speak. Whisper performs well for clean narration - the kind you'd record for a tutorial or demo. For final captions, a quick review pass is still worthwhile.
Does my audio get sent anywhere?
Never. Transcription runs entirely on your device using Whisper via CTranslate2. No audio leaves your machine - not during transcription, not ever.
Can I edit the transcript?
Not yet - but it's coming. The current release generates an SRT file alongside your export. Text-based editing - trimming by editing the transcript, removing filler words - is on the roadmap.
Up and running in seconds
One command. That's it. Works on Debian, Ubuntu, Fedora, Arch, and derivatives.
Not on a supported distro? — note: limited support on wlroots compositors (Sway, Hyprland).
Opening in your software manager...
If nothing happens, download the .flatpakref file instead.