Repository navigation
Releases: Shob3r/audio-replacer
Release list
5.0.3
Audio Replacer 5.0.3
Minor update to address security vulnerabilities and update app id.
Changes
- Updated all packages to their latest version. This should hopefully resolve most security vulnerabilities in the project
- Change app ID from
org.lemon-studios.audio-replacertodev.shob3r.audio-replacer
5.0.2
Audio Replacer 5.0.2
Security release to address several CVEs from Svelte's packages
5.0.1
Audio Replacer 5.0.1
This project is always going to be cursed with an immediate hotfix release after a major update, I swear
New Features
- Added an option to change the intensity of the noise suppression filter, if enabled
Removed Features
- Remove an unused total time with app open statistic because I couldn't figure out how to get it to work. Maybe in the future when I revisit the project....
Fixes
- Added type checking on any number input settings to prevent situations where invalid values would be in the settings
- Fix FFMpeg builds not building with the arnndn model, which is required for noise suppression
Full Changelog
5.0
Audio Replacer 5
The largest release of Audio Replacer is finally ready after 8 months of development.This version initially focused on simply porting the app to Tauri for performance improvements, but has evolved to be a huge update on a whole different scale compared to previous version updates
New Features
New UI
Everything in Audio Replacer 5 is UI designed specifically for Audio Replacer 5. No more pre-made components made by Microsoft to follow their design standards.
Multi-Project Support
Instead of there being one massive global project, Audio Replacer 5 stores projects in their own files, which can be loaded at any time from the home page. A history of all projects are visible in the home page. Custom pitch and effect filters have also moved to these custom project files, making them more modular than before
Visual Effect Editor
Instead of a clunky and hard-to-use text-based data editor, the new Effect Editor in Audio Replacer 5 is completely UI based, with simple and easy ways to add, modify, and delete effects. On top of that, you shouldn't need to reload Audio Replacer anymore to be able to select these effects. Additional safeguards have also been added, making sure that you don't insert invalid filters into the app
App Statistics
Audio Replacer can (optionally) track your usage of the app and display it on the Home page as your stats. Don't worry, stats stay on your device and can be deleted at any time
Linux Support
The switch from the Windows App SDK to Tauri has enabled me to create builds for Linux. Currently, only .deb and .rpm packages are available.
Changes
Improved App Performance
Switching to Tauri has allowed me to rewrite the most intensive parts of Audio Replacer in Rust, allowing for text transcription times of less than 1-2 seconds on the CPU and project load times of only a few seconds at most.
Custom Audio Replacer FFMpeg build
I have created a build of FFMpeg that has pretty much everything disabled except for audio filters and tools required to use them. This reduces the size of the binaries from 40Mb in Audio Replacer 4.3.2 to 2-4Mb in Audio Replacer 5, depending on platform.
Removed Features
I have only removed one feature in Audio Replacer 5: Vulkan/CUDA support for whisper text transcription. Due to the speedups, I believe it is unnecessary to support those runtimes anymore, as
- They are absolutely not needed
- App size would be inflated if they were to be included
This will be the last major update to Audio Replacer for a long while, unless I get an idea and want to update it. I will probably release minor and regular updates that may or may not add features and/or fixes (albeit likely quite infrequently), but major updates are complete for the time being. I believe this is the best and final form of Audio Replacer yet, and this will likely be the last major version update that will be used for the reason this project was created in the first place.
Full Changelog
4.3.2
Changes
- FFMpeg now bundled with Audio Replacer, no need for a download
- Implemented automatic migration steps for older version (no more manual command prompt entries)
- Changed speech transcription/whisper download option in settings to always allow download, not just when it's not already downloaded
4.3.1
4.3.1
Changes
- Replace setup complete check (.setupCompleted) with new hidden json option
Fixes
- Fix discord rich presence enabling some visual features relating to the recording page
- Fix low resolution app icon file (replaced 64x64 icon with 512x512 icon)
Changes in 4.3.0
New Features:
- New Home page. You are able to load the last opened project, load a different project, view patch notes, and see tips for the app
- Automatically loading projects has been removed, but it's one button click to load the previous project
- Add button to quick redirect to the Audio Replacer Wiki
- Add prompt to download updates instead of automatically downloading them
Changes
- Update app icon and installer splash
- Update text on some UI elements
- Replace Gyan.Dev FFMpeg with custom audio libraries only FFMpeg that reduces app size by 40mb
- You can learn more here
- Update Whisper.NET model to Q4_0-small, reducing model size by 100mb
- Update some project dependencies
- General app speed improvements
- Update Discord rich presence descriptions ("In the Data Editor" -> "Data Editor", etc.)
- Add Timeout to speech to text transcription based on what runtime is being used:
- CUDA: 3s
- Vulkan: 12.5s
- Cpu (With Avx instruction set): 20s
- Cpu (Without Avx instruction set): 35s
Removed Features:
- Removed option to remember last opened project. App will now always remember last selected project and it's a lot easier to load different projects from the home page
- Removed option to check for updates, as they are no longer forced upon you
- Removed option to toggle app converting input files to .wav as they are the preferred format of the application
What's Next
That's probably it. Audio Replacer is at a really good spot and I plan on working on other projects from now on. If i think of anything good to add or if bugs are discovered, I will provide bugfix patches, but I don't expect to work on this project this much anymore (Unless I find some worthwhile feature to integrate).
Full Changelog: 4.2.0...4.3.1
4.3
New Features:
- New Home page. You are able to load the last opened project, load a different project, view patch notes, and see tips for the app
- Automatically loading projects has been removed, but it's one button click to load the previous project
- Add button to quick redirect to the Audio Replacer Wiki
- Add prompt to download updates instead of automatically downloading them
Changes
- Update app icon and installer splash
- Update text on some UI elements
- Replace Gyan.Dev FFMpeg with custom audio libraries only FFMpeg that reduces app size by 40mb
- You can learn more here
- Update Whisper.NET model to Q4_0-small, reducing model size by 100mb
- Update some project dependencies
- General app speed improvements
- Update Discord rich presence descriptions ("In the Data Editor" -> "Data Editor", etc.)
- Add Timeout to speech to text transcription based on what runtime is being used:
- CUDA: 3s
- Vulkan: 12.5s
- Cpu (With Avx instruction set): 20s
- Cpu (Without Avx instruction set): 35s
Removed Features:
- Removed option to remember last opened project. App will now always remember last selected project and it's a lot easier to load different projects from the home page
- Removed option to check for updates, as they are no longer forced upon you
- Removed option to toggle app converting input files to .wav as they are the preferred format of the application
What's Next
That's probably it. Audio Replacer is at a really good spot and I plan on working on other projects from now on. If i think of anything good to add or if bugs are discovered, I will provide bugfix patches, but I don't expect to work on this project this much anymore (Unless I find some worthwhile feature to integrate).
Full Changelog: 4.2.0...4.3.0
4.2
New Features
- The custom pitch/effect dropdown menu has been replaced with a search bar to speed up finding the right options for larger custom data files
- Hotkeys for common tasks on the record page have been introduced
- Option to reinstall ffmpeg in settings (also technically works as an ffmpeg updater)
- New button in record page to switch between viewing the original and replaced audio
Changes
- Replace gyan.dev ffmpeg full build with gyan.dev ffmpeg essentials build (~60mb file size reduction)
- Introduce Release channels (Will be covered in its own section)
- Update icon for release logs page (now called "What's new") and move to bottom of the navigation tab
- Update credits tab in the settings page to include all project dependencies
- General performance improvements
Fixes
- Fix subdirectories of subdirectories not working properly with the app
- Fix issue where the extra data folder would not be created if not present
- Fix wrong icon being shown for theme selection in the setup window
- Fix no app icon being shown in alt+tab and win+tab view
- Fix issue where app would crash if no audio files are present in a project folder
- Remove debug file for a project dependency that was inflating app size by ~60mb
- Improve file searching speed in larger projects
Release Channels
Whisper.net's cuda runtime is over 300mb in size, which inflates the app size to beyond unreasonable for most users. for that reason, I have created two release "channels": "stable" and "stable-nvidia". Users who have Nvidia GPUs and want to take advantage of faster speech to text processing should download the Nvidia installer file from this release. Automatic updates should download from your respective channel.
What's Next
The next release (version 4.3) will be a optimization patch aimed at removing as many unneeded dependencies as possible, as well as generally making the app at least slightly better to use (Although there probably will be a few new features in the update regardless)
Full Changelog: 4.1.3...4.2.0