Releases

Added both norwegian languages

Added both norwegian languages

Implemented a statistics page and functionality that now enables the user to see how many words they Codictate, what their words per minute is and what their streak is.

Implemented a statistics page and functionality that now enables the user to see how many words they Codictate, what their words per minute is and what their streak is.

updated the compiled version of the mic recorder and the key listener on macOS.

updated the compiled version of the mic recorder and the key listener on macOS.

Fix some permissions logic to enable the user to select all of them so that none of them are blocking to each other inside of the onboarding.

Fix some permissions logic to enable the user to select all of them so that none of them are blocking to each other inside of the onboarding.

Translate mode was not enabled for all the models that was actually capable of doing translation, so now that has been fixed. Furthermore, we've now made it possible for users to disable the start and stop sounds for the transcription.

Translate mode was not enabled for all the models that was actually capable of doing translation, so now that has been fixed. Furthermore, we've now made it possible for users to disable the start and stop sounds for the transcription.

Fix windows issues

Fix windows issues

Codictate v0.0.47

Codictate v0.0.47-canary.1

I updated the model showcasing to now also include the RAM which is way more valuable for the user instead of the redundant disk showcase.

I updated the model showcasing to now also include the RAM which is way more valuable for the user instead of the redundant disk showcase.

Updated the accuracy and speed calculations to be proper and include the account for English only models and now we display the featured models that we are also displaying on our website so that they align. Now the featured models are based on the benchmarks that we have done.

Updated the accuracy and speed calculations to be proper and include the account for English only models and now we display the featured models that we are also displaying on our website so that they align. Now the featured models are based on the benchmarks that we have done.

Now the user is able to make informed decisions about all the whisper models that we can provide to them so that they can start using that in the app.

Now the user is able to make informed decisions about all the whisper models that we can provide to them so that they can start using that in the app.

Updated to now include benchmarks as well as fixing the parakeet model download size

Updated to now include benchmarks as well as fixing the parakeet model download size

Big big design change. Now we support light and dark mode. We have also changed a lot of the UI as well as implementing a big new feature called history mode. Now in history mode you are able to save transcripts to later copy and paste it. You are able to in transcripts to both save the audio and the transcript if you want to. or you can opt out related to the audio. A new Logo has also arrived!

Big big design change. Now we support light and dark mode. We have also changed a lot of the UI as well as implementing a big new feature called history mode. Now in history mode you are able to save transcripts to later copy and paste it. You are able to in transcripts to both save the audio and the transcript if you want to. or you can opt out related to the audio. A new Logo has also arrived!

Major design update. The whole UX of Codictate has been re-evaluated and remade to make it more accessible and more easily usable for the users. More Windows system support has also been added.

Major design update. The whole UX of Codictate has been re-evaluated and remade to make it more accessible and more easily usable for the users. More Windows system support has also been added.

updated to make sure that models can now successfully be downloaded without getting the problems that was caused by a simple mistake in one of the download scripts.

updated to make sure that models can now successfully be downloaded without getting the problems that was caused by a simple mistake in one of the download scripts.

Codictate v0.0.39

Codictate v0.0.39-canary.1

Codictate v0.0.38

Codictate v0.0.38-canary.1

Better windows support with proper icon. Smaller packages updates

Better windows support with proper icon. Smaller packages updates

updated the permissions to hopefully be more resilient based on users reporting not being able to accept accessibility permissions.

updated the permissions to hopefully be more resilient based on users reporting not being able to accept accessibility permissions.

Updated windows and linux to also have a logo

Updated windows and linux to also have a logo

Codictate 0.0.34

Codictate 0.0.34-canary.1

Big update to the format mode. Now we make use of llama.cpp and the Qwen 2.5 instruct model. This has been in place to make it more accessible for more operating systems as well as using a model that's better at following instructions.

Big update to the format mode. Now we make use of llama.cpp and the Qwen 2.5 instruct model. This has been in place to make it more accessible for more operating systems as well as using a model that's better at following instructions.

updated the dictionary commit approach

updated the dictionary commit approach

New feature called Dictionary. Dictionary enables users to correct things in the transcriptions they make and then Codictate will learn based on their corrections.

New feature called Dictionary. Dictionary enables users to correct things in the transcriptions they make and then Codictate will learn based on their corrections.

updated the formatting and the settings related to formatting, so it makes more sense and is easier to understand for the user. Furthermore, we have tried to update some of the system prompts to make the formatting more reliable. It's not completely there, but it's improved.

updated the formatting and the settings related to formatting, so it makes more sense and is easier to understand for the user. Furthermore, we have tried to update some of the system prompts to make the formatting more reliable. It's not completely there, but it's improved.

Fixed the duck audio manual input so that it actually adheres to what the slider says for the main audio and for the headphones as well as preventing the user from accidentally transcribing an old transcript when accidentally hitting the tap and hold shortcut.

Fixed the duck audio manual input so that it actually adheres to what the slider says for the main audio and for the headphones as well as preventing the user from accidentally transcribing an old transcript when accidentally hitting the tap and hold shortcut.

Fixed an issue with the indicator not being visible on full screen windows as well as rewriting it to a native component so it also now works with different window managers like aerospace.

Fixed an issue with the indicator not being visible on full screen windows as well as rewriting it to a native component so it also now works with different window managers like aerospace.

Implemented a new feature called Format mode. Format mode enables the user to format the output of their transcription based on the application they are in. In this first version it enables the user to format text in emails, slack, messages and documents.

Implemented a new feature called Format mode. Format mode enables the user to format the output of their transcription based on the application they are in. In this first version it enables the user to format text in emails, slack, messages and documents.

Major release to the looks of the application as well as a new model. We're now introducing the NVIDIA Parakeet model that enables the user to transcribe insanely fast. It comes at the cost of a bigger size, but that it's the fastest 1 we have yet. The limitation that it has is that it only supports around 25 languages, and it does not support translate mode. With this new model, we also introduced stream mode, which enables the user to speak and then in real time is able to see what they are saying.

Major release to the looks of the application as well as a new model. We're now introducing the NVIDIA Parakeet model that enables the user to transcribe insanely fast. It comes at the cost of a bigger size, but that it's the fastest 1 we have yet. The limitation that it has is that it only supports around 25 languages, and it does not support translate mode. With this new model, we also introduced stream mode, which enables the user to speak and then in real time is able to see what they are saying.

The application is now signed using Apple's certificates which will make it easier for users to actually download the application.

The application is now signed using Apple's certificates which will make it easier for users to actually download the application.

Hardening of the translatemode settings manipulation

Hardening of the translatemode settings manipulation

Config updates to better handle language selection, config, and storage

Config updates to better handle language selection, config, and storage

Hopefully fixed the translate problem

Hopefully fixed the translate problem

Updated Translate Mode config chnages to prevent default language overriding auto-detect

Updated Translate Mode config chnages to prevent default language overriding auto-detect

Updated the duck logic to duck the audio 100%

Updated the duck logic to duck the audio 100%

Fixed that the star dictation sound is muted too early.

Fixed that the star dictation sound is muted too early.

Now enables Coditate to lower audio when playing from speakers so music and podcast doesn't interfere with the audio wanting to be transcribed.

Now enables Coditate to lower audio when playing from speakers so music and podcast doesn't interfere with the audio wanting to be transcribed.

Added the indicator to the onboarding flow so the user can decide whether they want the indicator or not.

Added the indicator to the onboarding flow so the user can decide whether they want the indicator or not.

Fixed, so now the mission control is not affected by the indicator moving so far off the screen.

Fixed, so now the mission control is not affected by the indicator moving so far off the screen.

Big update, so now the user will be able to see a physical small indicator if they enable it in the settings. This is to make it more visually available for the user to know that they are actually speaking instead of only the sound cues.

Big update, so now the user will be able to see a physical small indicator if they enable it in the settings. This is to make it more visually available for the user to know that they are actually speaking instead of only the sound cues.

Major update to how the shortcuts are being used in Codictate. This includes the ability now to push to talk and then when you release it's automatically pasted. Now you also have the option to set up a secondary shortcut that only works with the push to talk.

Major update to how the shortcuts are being used in Codictate. This includes the ability now to push to talk and then when you release it's automatically pasted. Now you also have the option to set up a secondary shortcut that only works with the push to talk.

Updated the permissions logic along with an extra product onboarding screen that enables the user to try out the product for the first time.

Updated the permissions logic along with an extra product onboarding screen that enables the user to try out the product for the first time.

Updated the onboarding flow so that the permissions requests come one at a time.

Updated the onboarding flow so that the permissions requests come one at a time.

Updated some logic to make sure that when switching away from translate mode we are actually switching back to auto mode instead of staying in the default language.

Updated some logic to make sure that when switching away from translate mode we are actually switching back to auto mode instead of staying in the default language.

Fixed the tray logic for translate mode which was not doing the same as the Translate Mode button in the application

Fixed the tray logic for translate mode which was not doing the same as the Translate Mode button in the application

Big update that now enables the user to use translate mode. Translate mode enables the user to speak in their own language and then their dictation is automatically translated to English.

Big update that now enables the user to use translate mode. Translate mode enables the user to speak in their own language and then their dictation is automatically translated to English.

Better handling of the brand name in the output.

Better handling of the brand name in the output.

Added the ability to change the recording limit. So now there is 2 through 7 and 15 and 30 minutes. Also added a text explaining that this can cause the transcription to be slower and the recording will take up more disk space.

Added the ability to change the recording limit. So now there is 2 through 7 and 15 and 30 minutes. Also added a text explaining that this can cause the transcription to be slower and the recording will take up more disk space.

More robust handling of the Whisper language code.

More robust handling of the Whisper language code.

Changing of the transcription language is now accessible from the tray menu.

Changing of the transcription language is now accessible from the tray menu.

Language selector for better accuracy on harder or less spoken languages.

Language selector for better accuracy on harder or less spoken languages.

Update of application designed to match the brand that is being built.

Update of application designed to match the brand that is being built.

Updated the cancel dictation logic

Updated the cancel dictation logic

First release to the main repo. This includes a model that understands multiple languages. This is the start of the OpenSource era for Codictate

First release to the main repo. This includes a model that understands multiple languages. This is the start of the OpenSource era for Codictate