Hey Marshall! Cool to see this coming together, kudos for buildimg the tool you wish you had, thats the right reason to do things!
it seems like these “realtime meeting assistant / transcriber” services have taken a huge leap closer to being what I too have have often found myself wishing for. (Recently I gave Hedy AI a shot, very much in the same neighborhood functionally feels like)
out of curiosity, for Mimic’s Local Mode, whatre the tech specs required for a reasonable level of performance?
Matt, Thanks!
I just tried Hedy, same concept, also a great tool. It's todos are nice.
MimicScribe works well with any Apple silicon Mac so it'll feel snappy on an M1 with 8GB of RAM even. It uses Apple's on-device ML accelerator, the ANE. https://mimicscribe.app/docs/performance
I think the privacy story is super important, I like the use of local models as the fix here and understand that the output wouldn't be as good as frontier models. Perhaps look into venice for more private inference.
venice.ai? Ok nice I'll check it out. thanks. ya I added the openai compat endpoint, no brainer.
Looks great! Feature suggestion: would be great to plug in ollama for the AI parts. Not as great as a BYOK, but worth it to those who want to keep everything local
Hey thanks! Good call. I had limited success getting Qwen 3.5 9B working for some of the longer prompts that require lots of json output. I feel like completely on-device is so close to being usable for this stuff though. I should revisit this, actually.
running locally is great, but I would wonder about the system requirements (I love my mac air but ollama on 8GB with an Intel i5 probably would be a little too much for the little thing). But having a toggle option would be great
Right? I feel like local model support is something that has strong ideological appeal and might influence someones feeling towards the app, but when it comes down to it most people will probably just use a cloud model for larger tasks unless they have beastly hardware. It's like how I have an Android in part because I might one day flash the ROM.
Ollama just exposes models via an OpenAI compatible endpoint though (I'm pretty sure), so adding that standard is probably a good idea. The prompts are a bit tuned for Gemini. I'd have to test how much that matters.
Somehow I expected from the headline it identifies loudspeakers by their sound signature and got really curious :)
lol that'd be a trick. I'd have it purposely misidentify to cheaper brands to mess with people.