Changelog
Follow up on the latest improvements and updates.
RSS
v0.0.54 has been (re)released!
We’ve fixed the error from our last update and added some exciting new features.
Also, we realized we didn’t share what changed in version 53, so we’ll cover that in this changelog too!
[Built-In Templates]

- We’ve added built-in templates to help you save time and create higher-quality meeting notes more easily.
- You can select these templates directly from the settings modal and start using them right away.
- Pro Tip: The “General Meeting” template works great for most types of meetings :)
Want more template options? Tell us what you'd like to see!
[Obsidian Export (Beta)]
We heard from many of you that you'd like more integrations and sharing options.
In version 54, we’ve added Obsidian export (Beta), and email export is coming very soon!

To export to Obsidian, head to the Integrations tab in the settings modal and connect Hyprnote to your Obsidian app.
For more details, please refer to the guide in our docs.
[Template Selection After Recording Ends]

- Starting in version 53, we added the ability to select a template immediately after a recording ends via the stop recording modal.
- This makes it easier to apply the right template without navigating back and forth between the settings and the note editor.
That's it for v0.0.54! Thank you always for your support :)
new
v0.0.52
v0.0.52 is here!
We initially rolled out v0.0.51 last night but found some bugs related to enhancements with custom templates. We quickly fixed them and re-deployed v0.0.52. We apologize if any of you experienced any inconvenience due to the issue 😥
[Upgraded enhance button]
two major changes have been made to the enhance button.
First, if you're using a local LLM, you'll now see a progress bar during the enhancement process, so it's clear what the AI is doing.

Second, when re-enhancing a note, you can easily choose which template to use from the re-enhance button.

[Basic speaker diarization]
A lot of our users have been asking for this feature, so we built it!
While it’s not perfect (it can’t yet recognize and categorize different individual voices), it can now differentiate between sound from your microphone (what you’re saying) and sound from your system (what others in your video call are likely saying), and categorize them in the transcript.
You’ll see the transcript with speakers separated once you stop your recording.

[Miscellaneous]
- Fixed notification related errors
new
v0.0.50
We rolled out a lot of new features in v0.0.50!
[Custom template editor]
A feature many of you have requested is finally here! You can now create your own custom templates in the settings modal.

To get started, click the ‘+’ button at the top right to create a new template. The editor is simple and intuitive. You can even choose a custom emoji for each template to keep things organized.

After saving your template, return to the template list and select the one you’d like to use. When a template is highlighted with a blue outline, our AI will automatically apply it when enhancing your notes.

We’re also working on a way to let you switch templates directly from the editor or note area without opening settings. This improvement will be available in the next few days—stay tuned!
[Command palette]
We’ve improved search with the new command palette.
Press Cmd+K (or click the search bar at the top) to open it. From there, you can quickly search across all your notes, people, and organizations in one place.

[We want to hear from you]
Your experience with Hyprnote is our top priority. To make it easier for you to connect with us, we’ve added more ways to get in touch directly in the app.
Just click “Talk to Founders” or tap the version information in the settings popover to share feedback, ask questions, or learn more about what we’re working on. We’d love to hear from you!

[Improved transcription area usability]
We’ve made recording smoother by removing unnecessary scrolling in the transcription area during meetings. This should make transcriptions much easier to follow in real time.
[Fetch upcoming events]
You’ll now see your upcoming events correctly in the left sidebar alongside your notes and sessions. No more errors!
We also fixed a number of internal bugs to improve overall stability and performance.
new
v0.0.49
- Whisper memory issue fixed. (Unexpected crashing introduced in v0.0.48)
- Upcoming events list fixed (thanks @plyght )
- Remove duplicated Sentry script in frontend
- Add AEC(Acoustic Echo Cancelation). Transcript & recorded audio should be much better when you use Hyprnote without Headphone.
We also added some one-time survey(thanks @Deokhaeng Lee ) to better understand our users. Thank you for the support.
new
v0.0.48
- Export to PDF: You can now export your enhanced note as a nicely formatted PDF by pressing the 'share button' on the top left corner of the editor area!
- New Search/Replace Bar UI: You can now easily activate search/replace bar for the transcription with cmd+f.
- HOTFIX: Tag generation errors are fixed so that it actually works 🙂
new
v0.0.47
- GPU Fallback: If your GPU doesn't work for llama, Hyprnote will now transition into using CPU instead.
- Updated Token Limits: we have increased the token limit for enhancements, so now you'll be able to enhance and summarize longer meeting transcripts!
- Custom LLM Endpoint: we fixed some errors so that setting the custom LLM endpoint experience feels smoother and actually works! (you can always change enhance LLM by going to Settings -> AI -> Enhancing)
new
v0.0.45
- HOTFIX: invalid metadata passed to LLM provider. (only relevent for custom endpoint users. Some provider like OpenRouter works though)
- Native support for Thinking models. @bhubbb made Dethinker(https://github.com/bhubbb/dethinker) to workaround this before. Now works out-of-the-box.
new
v0.0.44
- Generate title if enhancing is done but title is empty
- Support custom endpoint that does not have /models endpoint
- Right panel UI update - Search + Responsive
+ 🫡
We're cooking hypr-llm which is trained for meeting summarization and related tasks.
new
v0.0.43
- fix crashing when copy note editor
- Better formatting on copy for both note/transcript editor
- fix invalid utf8 error in whisper inference
- bump whisper/llama.cpp version
new
v0.0.42
- Transcript Editor performance fix. No more lagging.
- Prevent default drag-n-drop handler (We will later support proper file uploading later)
- Fix markdown formatting problem when using custom api
- Other small UI improvements (e.g traffic light position)
Load More
→