Option to upload an audio file for local transcription

[themaximehardy]

Hi everyone 👋

First off, thanks a lot for building Handy!! It’s such a great tool! I love how simple and fast it is, and the fact that everything runs locally is a huge plus.

I had a small feature idea that could make it even more useful:
it would be amazing if we could upload an audio file (like .mp3, .wav, .m4a, etc.) and get a full transcription directly inside Handy – still 100% offline, of course.

Right now, Handy works perfectly for live speech-to-text through the microphone, but sometimes I already have a recording (for example, a voice memo, interview, or meeting) that I’d love to transcribe without using any online service.

A simple “Upload audio file” button or drag-and-drop area would be great. The app could then process it locally and show the text (or let us copy it).

Anyway, just wanted to share the idea – it feels like a natural next step for Handy, and I think a lot of people would use it.
Thanks again for all the work you’ve put into this project 🙏

[PhunkyBob]

+1 for this feature
And if it could do diarization, it would be top-notch!

[oblab]

+1 too

[NearlyUnique]

If as well as or instead of "upload/import audio" a command line switch was possible

handy my-audio.mp3 --output my-transcript.txt

[olejsc]

I've made a PR with a PoC for this feature #371 (only tested on Windows). Diarization is not included in it yet, as I didnt want to drag in any new dependencies (and also because I have no clue how that works..)

[cjpais]

Overall the main feature of local transcription will be coming soon

[lbrese]

same

[koloved]

i need that man, openwhispr has it but i wanna use handy

[cjpais]

Please give me time. It will happen.

[iTzFaisal]

Thanks @cjpais for your hard work. I love Handy !

[lukefilestalker]

that feature would be super helpful :)

[longyuxi]

This feature will also enable transcription on mobile phones. Once this feature is done one can imagine using a widget or shortcut on their phone to record a message and send the recording to a locally running instance of Handy. This beats existing solutions (e.g. using a shortcut to send to the whisper API or using the clunky Wispr Flow app) by a lot in extensibility and privacy. Happy to contribute a PR after the above feature is implemented.

[javi22020]

Has this been done? Else tell me, I'd like to contribute (it's a feature I need)