Privacy first, open source dictation and meeting recordings without a bot. 100% free with local models or bring your own key.

sf
Joined April 2024
the first (and only?) open source dictation app 🤭
the openwhispr iOS app is out of beta and we just crossed 9,000 users! (we filmed the launch 2 weeks ago😅) as always, it's 100% open source and free with on-device models or when you bring your own api key give it a try and let us know what you think!
4
370
OpenWhispr retweeted
Will this be the first open source iOS dictation app? 🤔 this is 100% on device and free 🤘
2
5
23
851
OpenWhispr retweeted
The @openwhispr team (SR007) is hiring a found eng + a social/brand creator! Been a pleasure see @gabrielste1n build a fully private voice-to-text solution. If you are big on private alternatives + a company with clear traction...you should apply! Careers: openwhispr.com/careers
After months of work, OpenWhispr is now SOC 2 Type II and HIPAA compliant, and ISO 27001 certified 🔒 Thousands of people trust us with their conversations, ideas, and work. That’s a responsibility we take seriously. Giving you control over your data has been central to OpenWhispr from day one, and we’re continuing to invest in keeping it secure as we grow. Thanks to our OSS friends at @compai for the support!
3
2
29
4,997
After months of work, OpenWhispr is now SOC 2 Type II and HIPAA compliant, and ISO 27001 certified 🔒 Thousands of people trust us with their conversations, ideas, and work. That’s a responsibility we take seriously. Giving you control over your data has been central to OpenWhispr from day one, and we’re continuing to invest in keeping it secure as we grow. Thanks to our OSS friends at @compai for the support!
1
3
25
5,541
OpenWhispr v1.10.0 ships with a speed upgrade! Orukeet wins 61 of 74 benchmark splits; Parakeet wins 13. Hoid's latest OpenWhispr PR: 171 → 110 ms for Orukeet (36% lower warm median). Apple M5, 11-second clip, before/after optimization.
1
1
14
719
OpenWhispr v1.10.0 ships with a speed upgrade! Orukeet wins 61 of 74 benchmark splits; Parakeet wins 13. Hoid's latest OpenWhispr PR: 171 → 110 ms for Orukeet (36% lower warm median). Apple M5, 11-second clip, before/after optimization.
step aside @nvidia, there's a new top dog in speech models: orukeet. trained by @OrukLabs, optimized by @hoid_ai, and available on day zero in @openwhispr (100k+ downloads). 25 languages. Open weights + code.
5
5
2
23
10,637
OpenWhispr retweeted
We just crossed 100k downloads of OpenWhispr! 🤯 This feels like a good time for a reminder of why we’re building it. Your ideas and conversations are becoming more valuable, while software is becoming easier to build than ever. But that creates some worrying incentives for existing software companies: use your proprietary data to train better AI models to stay competitive, and lock that data behind walled gardens to keep you from leaving. We think that’s pretty shitty. You should stay because a product is useful, not because your data is stuck there. If billions of people are going to feel comfortable sharing their most private information when talking to AI, and organizations are going to trust it with their work, we have to earn that trust. For us, that starts with being fully transparent about how your data is handled, and giving you full control over it. That’s what we’re building with OpenWhispr. You can run models on your own device, use your own cloud provider, or use our cloud service. We’re not anti-cloud. We just think it should be your choice. We’re still super early, but I’m genuinely excited to bring you along for more of this. This is the first in a series of videos sharing the new features we’re building, what we’re figuring out, and the team behind it. And if you’re still here reading, we’re officially hiring for two really important roles: - Founding Engineering Lead - Social / Brand / Growth Creator You can find more info about both on our website!
42
14
7
151
14,356
OpenWhispr is a free, open-source app that turns your voice into text on your computer, and it can work entirely offline so your audio never leaves your device. Explore it here: osp.fyi/open-whispr
7
116
1,087
54,331
OpenWhispr retweeted
So cool to see OpenWhispr trending on GitHub this week! If you haven’t already - a star actually really goes a long way! github.com/OpenWhispr/openwh…
7
2
22
912
OpenWhispr retweeted
Replying to @openwhispr
@openwhispr is so freaking good! Even on a 24GB M4 Macbook air with local models for everything it works FLAWLESSLY and it's so convenient! I'm regretting not using it sooner.
1
1
3
104
OpenWhispr retweeted
Replying to @AggyAbhishek
@openwhispr for life
1
2
160
OpenWhispr retweeted
Stoked to share that I’ve joined @speedrun 007 as a solo founder to keep building @openwhispr It’s becoming clear that voice is the natural interface for AI. When you talk instead of type, AI gets the nuance, context, and intent it needs to turn your thoughts into action. But those conversations contain something incredibly valuable: your ideas, your knowledge, and your intellectual property. We believe individuals and organizations should have full ownership and control over that context — not closed platforms. I’m already getting so much value from being part of the program, and I’m super grateful to have the support of @JoshLu @far33d @andrewchen @_CallMeMacy @tabishgilani @justmazer @thetalentcarver @nazzari @Chen and the whole speedrun team And a special thank you to @kobyjconrad and @shaiunterslak for making the introduction to the speedrun team in the first place. Really appreciate you guys ❤️
51
6
1
172
9,379
OpenWhispr retweeted
This is one of our biggest use cases of @openwhispr - multilingual users can speak in their native language and it gets translated to English (or any other language they want)
Imagine Slack with built-in translation so everyone can communicate in their native language. No more thinking in English first or rewriting to sound natural. I just want to type: “Che, vieja, ajustame ese loguito, dale pa.”
1
3
593
Shoutout to @openwhispr it's pretty rad, especially if you run both the speech-to-text and the language model to clean the transcripts up locally. glad i got this 128gb unified memory
2
1
4
1,028
OpenWhispr retweeted
If you're not using WisprFlow or any other STT solution to write your prompts, you're missing out. I can get so much more detail in by just talking instead of typing - which gets me better responses from the model. I personally use OpenWhispr (@openwhispr) by @gabrielste1n. The best open source solution for this in my opinion.
5
2
1
17
2,857
Confession: I was on @WisprFlow for 6 months, kept hitting the free limit, and was juggling multiple accounts to keep going. Just switched back to @openwhispr. Barely any difference. Free Grok API key = ~2000 req/day, zero cost. Open source. Your key, your data.
1
2
580
Here's a quick overview and demo of Openwhispr: youtube.com/watch?v=XUi7iNyH…
For those looking into open source and BYOK alternatives to Wisprflow, meet OpenWhispr openwhispr.com/ - Can also have it join Zoom meetings for you and get notes, use local LLMs if you prefer, and keeps your data private. Also 2000/words per week (not per month) for free
1
4
993