Feedback
Roadmap
Changelog

Have something to say?

Tell us how we could make the product more useful to you.

Completed

Shortcut creation bug

Always seems to be an about shortcuts from me. 😁 Depending on how the keys are pressed when recording a shortcut, the shortcut will vary, which is very non-standard on macOS, don’t know about Windows. For example: If I record a shortcut by pressing Option and then W the shortcut is Option W. However, if I record the shortcut by pressing Option and W at the same time, the shortcut becomes Option ∑ which is going to fail when used. Because Vowen is adding the key press immediately, this what allows the user to release the first key, and then press the second key. What should happen is the shortcut is only recorded when all the keys are pressed at the same time. Also, and this is for consistency on macOS, the order should always be shown as Modifier + Key, never Key + Modifier.

Simon 1 day ago

2

Planned

Create Custom Utilities

Loving some of the utilities so far. There are gaps that this doesn’t cover for me. Community or local custom utilities would be really nice. Rough idea: Name, icon, trigger phrase Action: shell command, script, or AI prompt Variables like {file} / {files} substituted from the current selection Optional JSON export/import so people can share them

simp4_jesus 2 days ago

2

Completed

Partially downloaded models do not throw errors

Ran update while downloading model, led to model not fully downloading, but UI indicated it was fully downloaded. Model was only 37MB on disk (when should have been ~1.5GB). Only 3 out of 947 tensors were loaded and the wisper server kept restarting. On restart, I got a green check, which led me to believe the restart fixed the issue, but obviously it did not. Manually deleted model and was able to re-download with download link next to model (but UI still had a green check mark next to the model, which is confusing in my opinion). Request: If the model is not fully downloaded or corrupted, the system detects this, displays a red ! next to the model on the models tab, doesn’t allow selection of that model, and allows for re-download of the model.

Colin Fish 4 days ago

2

Hyperkey and Double Tap

So we ask for more keyboard shortcuts, we get them, and we’re still not satisfied. 😆 I use the app Hyperkey to assign the Right-Option key to function as control-option-command, so for example, I have an app that bring it’s floating window to the front with the shortcut Control-Option-Command-F, but I just have to press Right-Option-F. Sounds complicated, but it’s not. I’ve a bunch more assigned this way by pressing Right-Option-Key, and it works really well. I also have installed an app called Double Tap, which I’ve configured to trigger Control-Option-Command-F when I double tap Left-Option. None of the methods work with Vowen, while other Mac apps don’t have a problem with this setup. Is Vowen using a non-standard key detection method?

Simon 8 days ago

Completed

Allow Switching to Handsfree Mode Mid-Recording

Currently, in Vowen, handsfree mode can only be activated at the start of a recording. If I begin with the transcription hotkey, I cannot switch to handsfree without stopping and starting a new recording. Coming from Wisperflow, pressing the Fn key and then the bar button allowed seamless transition to handsfree mid-session. Adding similar functionality in Vowen would improve workflow and flexibility. Video Repro

Xavier 8 days ago

3

Completed

AssemblyAI API Error

Issue: AssemblyAI API Transcription Failure Root Cause: The transcription request is using a deprecated parameter. The error message indicates that the speech_model field is no longer supported by the AssemblyAI API. Technical Details: Service: AssemblyAI Error Message: speech_model is deprecated. Use "speech_models" instead. (Note: The error text in the screenshot contains a likely typo, but the intent is a field name change). Documentation Reference: https://www.assemblyai.com/docs/pre-recorded-audio/select-the-speech-model Required Action: Update the API request payload to replace the deprecated speech_model key with the current supported parameter (likely speech_model with updated values like best or nano, or the plural speech_models if using multiple models). The developer should verify the exact key name and valid string values against the linked documentation.

sjanett 10 days ago

2

Planned

add "start this prompt before showing text"

make it possible to let user type the prompt to be run after voice is recorded, for example, i need to run a token optimizer before every instruction so claude dont get lost in my thinking and just execute. so it would be something like: 1- voice recordining 2- prompt modifying the outcome of the recording 3- directly showing the edited text. SYSTEM INSTRUCTIONS: Technical Task Optimizer (Voice-to-Markdown) Role: You are an expert Technical Project Manager and Prompt Engineer. Your sole purpose is to receive raw voice transcripts (which are often messy, verbose, or conversational) and transform them into precise, Token-Optimized Technical Task Briefs. Processing Rules: Zero Noise: Completely eliminate hesitations, pleasantries, repetitions, rants, or filler words (e.g., "like", "basically", "so", "I mean"). Technical Extraction: Immediately identify the Target (where to act) and the Action (what needs to be done). Information Density (Token Saving): Use short, direct sentences and technical vocabulary. No wasted words. Exclusive Output: ALWAYS AND ONLY respond with a Markdown code block. Do not add greetings, confirmations, introductions, or comments outside the code block. Mandatory Output Template: # TASK: [Short & Descriptive Title] ## 1. [Component or Section Name] - **Target:** [Specific Page/Element] - **Action:** [Clear and direct instruction] - **Specifics:** [Any exact color codes, percentages, or exact data requested]

Marco Munich 14 days ago

In Progress

User selected post-process with AI

Hi, I’ve tried sooo many transcription apps that it’s getting out of hand. I’ve tried Handy, Handless, Pindrop, Spokenly, FluidVoice, Pipit, WhisperShortcut, Ottex, Snaply, to name just a few. But I keep coming back to Vowen. I will admit that some of the ones named above have so very nice features not presently available in Vowen, but none are dealbreakers for me. Once again, the app is excellent, and congratulations. So, to the feature request, which I’ve not seen any other app offer. Firstly, create additional Enhance transcription with AI prompts. The could be Clean up, Rewrite, Make markdown note, Email, and so on. Next, once the model transcribes the text, the user is presented with these options, which, when one is selected, then passes the transcription to the AI model to enhance using the respective prompt. This way, the user doesn’t need to decide which prompt to use first, they just speak, and then select what to do with it. Now, I did make a similar-sounding feature request about triggering a Shortcut after transcription to automate what to do with the text. For example, just speak, without first selecting an app to paste the text into, and once the Shortcut is triggered, it could grab the text (clipboard) and present to the user options what to do with it, such as adding it to an existing note. Or using Private Cloud Compute to enhance. I think these two features could co-exist. Thank you.

Simon 16 days ago

4

Completed

use default model as parakeet blazing fast. its very accurate and amazingly fast.

fluid voice have this model as default chack their app and model classification its much better so take inspiration from them. you app is way ahead of other in terms of functionalities like workflow and all. also workflow can be enhanced a lot by adding script kind of mode. i can simply select script file and on voice trigger it will run that script file. also one last thing on transcribing load dont show app logo its kind of big instead it can be some minimal smooth animation

Aman Umrao 20 days ago

2

Completed

Memory context feature says AI is not configured

Hello, I added a note to take into account, in the memory context tab, but when saving, it complains with the error: “File added but AI is not configured. Enable AI in settings to use memory context.” I triple checked, my AI provider is correctly configured (Custom API/OpenAI-compatible to a remote LLama.cpp/swap instance) and works normally. Is there any “hidden” option I need to use or maybe it doesn’t work with Custom API selected?

Luc Stepniewski 21 days ago

2

Completed

Paragraph splitting, auto-listing, and minor text corrections

Coming from Wisper Flow,. A few of the things that I really appreciated that I was hoping could be included in here are: 1) splitting up long messages into separate paragraphs automatically 2) resolving very minor each mistakes in an automated way. So for example if somebody was to say “3 o’clock, no wait, 4 o’clock”, it would be great for the 3 o’clock to not be there. 3) Adding an ability natively to create lists when somebody is listing something. I think some of this is doable with the own customization settings, but it'd be really great if this was somewhat native. I found that when I was playing with tones that it would substantially change what I was actually saying at times, which wasn't ideal. Overall, I really love this app. Amazing work to the Vowen team!

jolystep 24 days ago

5