The dictionary limit is the main thing holding the free tier back I'm a Vowen user, and the 50-word dictionary cap is honestly the only free-tier restriction that gets in my way. I'd happily see a much larger dictionary as a paid feature - it would be genuinely valuable for me and several people around me. For reference, wisprflow.ai lets you add thousands of words and tunes the system around them (optimal LLM prompts, etc.). I've set up a small model that runs locally on my laptop and proofreads my transcribed text. It comfortably handles far more than 50 words using the language model alone - I'd estimate 500 - 1000 words without trouble on my hardware, and I'd expect the same via API requests. So the technical ceiling is much higher than the current limit suggests. The feature I'd really love: personal fine-tuning What would be amazing is a way to fine-tune the model specifically for me - for my voice and my vocabulary. If there's any way I can help test or contribute to this, I'm in. Feedback: please don't paywall the system instruction on the custom API Locking the LLM system instruction behind a paywall doesn't make much sense. It's trivial to work around - with LM Studio, Llama, or any other provider, I can set a system instruction at the provider's own level when connecting the model. So a paid system instruction just pushes me to configure it elsewhere. Given that it used to be free, the better move is the opposite: keep the system instruction as a free feature.

Nikolai Vysotskyi 1 day ago

1

Vowen 0.4.3 on Windows starts and immediately exits with no visible UI unless launched with SwiftShader flags. Normal launch reaches app initialization but Chromium/Electron GPU process crashes repeatedly, then exits with: GPU process exited unexpectedly: exit_code=-2147483645 FATAL: content\browser\gpu\gpu_data_manager_impl_private.cc:415 GPU process isn't usable. Goodbye. Workaround that makes it launch: Vowen.exe --use-gl=swiftshader --use-angle=swiftshader --disable-gpu-sandbox Plain --disable-gpu did not fix it. The app works with SwiftShader.

Stef 2 days ago

1

After installing a new update, both my Pro license key and my OpenRouter API key were gone and had to be re-entered manually. Steps to reproduce: Purchase a Pro license key Install a new update License key is no longer saved and must be re-entered OpenRouter API key is also no longer saved and must be re-entered Expected behavior: License key and API keys should persist across updates.

Jo 4 days ago

5

Appreciate the app, and get that at some point, your work should be rewarded. Adding new features and make them paid features is ok. However, start asking to pay for existing features (such as AI customization) is not cool.

Xavier 7 days ago

3

I'd like to change the folder where the recordings are currently saved

David Muñoz 8 days ago

1

A nice-to-have would be to include more sections of the app in the menubar item. Why not include Notes, Tones, Workflows, and so on? In fact, you could just add all the sections for quick access.

Simon 17 days ago

1

Any chance of using an alternative to WebM recordings on macOS, as it’s not natively supported? While there are ways to play the recording back – Chrome or VLC, for example – it’s still not native. Not a major issue, just something to think about for the future. Thanks

Simon 17 days ago

I’ve only started to use the Notes feature, and it works great. Not that it seems to affect the transcription, but I noticed when I play back the saved .webm file it has a bit of an echo on the parts of the audio coming from the meeting screen audio. The audio from the built-in mics is fine. Like I said, the transcript still works, it’s just listening back to the audio is hard on the ears after a while. Thanks macOS 26 Vowen 0.3.1

Simon 17 days ago

1

I’d like to suggest adding the ability to transcribe files in a queue (batch processing). Right now, if I need to process multiple files — for example, three — I have to manually wait for the first one to finish, then start the second, and then the third. It would be much more convenient to be able to add multiple files to a queue and have them processed automatically. It would also be great to receive system notifications when each file is completed, as well as when the entire queue has finished. And thank you so much for this great app and for the work you’re doing — it’s really appreciated! 🔥 🙏

Georgii Vlasov 18 days ago

1

On Windows, I've noticed the app sits consistently around 300MB of RAM, sometimes a bit more, sometimes a bit less. I've used other similar tools in the past that hovered around 30MB or so, which made me wonder if there's room for optimization on that front. I also noticed my MacBook Air M1 running warmer than usual during the same workflow I normally do. I can't say for certain Vowen is responsible, but connecting the dots with what I saw on Windows, it felt worth mentioning — especially on a fanless machine where thermals matter a bit more. I completely understand that a tool with this many features has a certain overhead, and I'm not expecting it to be ultra-lightweight. That said, I was wondering if any resource optimizations are on the roadmap — things like a more progressive resource usage model, or the ability to put certain features into standby when they're not actively being used, so the app has a lighter footprint during idle or low-usage moments. Just to add context: I'm using cloud APIs only (Groq and OpenAI) — no local models running on my side — so the footprint isn't coming from inference. Thanks as always for the great work

dimezio 19 days ago

1

Hi Vowen team, I wanted to mention something I've noticed a few times, not sure if it's a bug or something I'm misconfiguring. I'm using Vowen with Groq and OpenAI o4 mini for transcription, with a specific language set in the preferences (not "auto detect"). Most of the time it works fine, but occasionally the transcription switches to a completely different language — I've seen Cyrillic pop up a couple of times, and once or twice something else entirely. It tends to happen in two situations: With short phrases, where it seems like there isn't enough audio context to stay on track Randomly in the middle of longer transcriptions — so the first part comes out correctly, and then a chunk somewhere in the middle suddenly switches language before going back to normal It's not super frequent, but noticeable enough that I figured it was worth flagging. My guess is the underlying model might be doing its own language detection at some point regardless of the setting, but I'm not sure if that's by design or not. If there's something I should configure differently, happy to try. Otherwise, hopefully this is useful feedback!

dimezio 19 days ago

1

Hi Vowen team, First off, I really like your project, it's really useful. I wanted to share a couple of ideas that would make it even more powerful: 1. Auto-learn from manual corrections When a word gets transcribed incorrectly and I fix it by hand in the text, it would be great if the app noticed that and silently added the correction to my custom dictionary in the background. So the next time I say the same thing, it already knows what I mean - no extra steps. For example, I often say "DeepL" (the translation tool), and Vowen transcribes it as something like "dipl". If it learned from that first correction, I'd never have to fix it again. 2. Bulk import words via .txt or .csv file Right now, adding technical or niche vocabulary one word at a time can be pretty tedious. It would be really handy to be able to upload a plain text or CSV file with a list of words to add to the dictionary all at once. This would be especially useful for people who work in specific fields - for example, developers who regularly say things like MySQL, PHP, Supabase, or prompt engineering terms, or really anyone working in a niche where standard speech recognition tends to struggle. Being able to prepare a word list in advance and import it in one shot would save a lot of time. Thanks for considering it, and keep up the great work!

dimezio 19 days ago

2

Feature Request: Start automatically when a meeting is detected in a specific application (e.g., Teams). I currently use the Plaud desktop app for this functionality. Additionally, it would be great to have a small icon indicating that a meeting is being recorded, located in the system tray rather than on the screen.

AG 19 days ago

2

I'm using your app in Slack a lot and it's always complicated to attack someone and have the “@” symbol. I've tried adding dozens of threads to replace different ones but saying at and name or “arobase” and name is not working very well, I always need to correct it.

Xavier 19 days ago

2

Hey guys, I really love your application. Thank you so much for keeping it free. I think the next big feature That would really help me is text to speech.Because if I would be able to mark a text and transform it into a nice sounding AI voice, that would be so helpful when consuming content and stuff. Because the built-in text-to-speech features in the operating systems are not really usable because they sound really really awful.

mtietz 20 days ago

Thank so much - awesome app.

_marcusaurelius 21 days ago

3

One of my shortcut triggers is Option Shift. On macOS, Option Shift Command V is Paste and Match Style, something I use often. Shouldn’t Vowen detect that the user is pressing additional keys beyond the trigger shortcut and prevent the transcription from activating?

Simon 21 days ago

2

Feedback 1: confirm custom instruction deletion when “Clear” is pressed Feedback 2: If a user navigates away from the configuration screen or closes the Vowen settings (or the application) altogether while they have unsaved changes to the custom, trigger a warning and confirmation message before the user navigates away or exits out.

sugarycat 24 days ago

1

Hi! I have been trying to combine keyboard shortcuts with mouse button 4 and mouse button 5, but it doesn't seem to register correctly. It records the mapping but it doesn't get applied.

oasas 25 days ago

Right now every transcription is sent to the LLM for post-processing regardless of the length and what you are trying to do. What would be great is to have some kind of mechanism to indicate if I need that post processing or not. In many cases if you’re just trying to dictate a couple of sentences, whisper is perfectly good for this, it’s only when I might be talking for longer (or waffling about a bit while I’m speaking and thinking out loud) that having that post-processing capability is super handy for clean up.

ajfisher 26 days ago

1