sipjca

You literally just compared a laptop running arch to a mac. You’re not the target audience lmao

Picked up a T14s in Shenzhen for ~$250 US and it’s a screamer. Best thing for sure

who was bought by openai

Great feedback :) also support for the v2 versions of the moonshine models should be out today!

There's an open PR in the repo which will be merged which adds this support. Post processing is an optional feature if you want to use it, and when using it, end to end latency can still be under 3 seconds easily

Not currently

okay... so I cannot get this to run on my mac. maybe something with the burn kernels for quantized?

will report a GitHub issue

I'm looking into porting this into transcribe-rs so handy can use it.

The first cut will probably not be a streaming implementation

thanks for your contribution :)

this is hilarious, person who barely uses service says he will leave the service, what a concept

I believe sensevoice, I’ll hopefully be implementing it soon enough

Tauri

It’s in the debug menu right now (ctrl/cmd+shift+d)

Yeah like the other commenters mentioned, using Bluetooth devices does not work super well at the moment. Hopefully I’ll have a fix at some point. There’s just some time over bluetooth to negotiate the connection and everything, and the app doesn’t do a good job showing this at all right now

On a Mac I definitely recommend using the internal mic even if wearing airpods

I’d just try it and fork handy if it doesn’t work how you want :)

It’s all local

It does unload it, and actually might be a good default for most people as the model loading does happen in the background as soon as you hit the key

Okay so it's more directly text replacements

https://github.com/cjpais/Handy/actions/runs/21025848728

There is also LLM post processing which can do this, and the built in dictionary feature

Handy can post process with LLMs too! It’s just currently hidden behind a debug menu as an alpha feature (ctrl/cmd+shift+d)

I totally agree with you and largely what you’re describing is one of the reasons I made Handy open source. I really want to see something like this and see someone go experiment with making it happen. I did hear some people playing with using some small local models (moondream, qwen) to get some more context of the computer itself

I initially had a ton of keyboard shortcuts in handy for myself when I had a broken finger and was in a cast. It let me play with the simplest form of this contextual thing, as shortcuts could effectively be mapped to certain apps with very clear uses cases

It doesn’t! Just makes it more accessible to more people I feel. There’s a cli version for Mac which I wrote first handy-cli

There’s a PR for this which will be pulled in soon enough, I can kick off a build of the PR if you want to download a pre release version

What microphone are you using?

Please send me a crash log!

How many consumers (not business) are genuinely using frontier models? You think OpenAI and Anthropic will forever serve the most intelligent models to free users? Heck they don’t already

Efficiency gains exist and likely will continue, as well as hardware generally accelerating, as software and hardware starts to become co-optimized. This will take time no doubt but 10-15 years is hilariously long in this world. The iPhone has barely been out that long

And to be clear I think the other arguments are valid I just think the timeline is out of whack

> It may be possible to run a "good enough" model on consumer hardware eventually

10-15 years?!!!! What is the definition of good enough? Qwen3 8B or A30B are quite capable models which run on a lot of hardware even today. SOTA is not just getting bigger, it's also getting more intelligence and running it more efficiently. There have been massive gains in intelligence at the smaller model sizes. It is just highly task dependent. Arguably some of these models are "good enough" already, and the level of intelligence and instruction following is much better from even 1 year ago. Sure not Opus 4.5 level, but still much could be done without that level of intelligence.

this is true, but linux desktop is still fragmented hell and will be for at least a few more years

yes, it works, I would recommend it even if it's a bit awkward at times

Yeah.. It's pretty damn hard to get working at all, and the systems it does seem to work on is quite limited as well. And also the developers of Tauri largely have seemed to have given up on Linux (which to some degree I understand, it represents a tiny amount of users for 99% of the problems due to the massive diversity in the ecosystem of Linux DE)

I fight with GitHub Actions on this all the time, and damn near think about running dedicated instances just to build from.

I've also had some degree of success with Tauri but not without pains. But to be fair most of the pains have come from the state of Linux DE (it's absolute hell if you ask me)

For the most part, things just work on MacOS (Windows I don't use much, but I don't get that many bug reports from Windows, so it must work alright. I would guess I have 50k Windows users?).

I don't have any experience with Electron, but in many ways I assume it probably is much more robust of an experience than Tauri. A lot of stuff has felt shaky to me, yet it does work. It's just not always a confidence inspiring experience. But again, maybe this is just the state of cross platform experiences generally? I don't know because I don't have enough experience to say

Source, I'm the author of: http://github.com/cjpais/Handy