Hi, i finally feel confident enough to announce Audionaut. It's been 3-4 years in the making and my initial motivation was to edit my own multi-channel recordings with the old Sound Designer II workflow (create regions, drop to a playlist, export playlist, done). It got a little bigger now and the most current feature is the agent editing, which is imho very useful since you have the UI which always provides transparency.
To try the agent part with Claude Code (app and Node 18+ installed):
claude mcp add audionaut -- npx -y audionaut-mcp
Edits land in the open project as one undo step each, and saving stays with you. On macOS the app is sandboxed, so keep projects in ~/Music.
My use case is pretty unique I guess and not something you built audionaut for. I do small live gigs in a local place where I sometimes need to operate lights, screen and play music at the same time and need something I can cue things to multiple software/devices through osc, websocket, tcp during music playback. I know there is already software for that but I didn't find anything I could run and quickly operate during the rehearsal before the event.
I was just wondering if you planed to add some features like markers or cue for integrations.
Have you ever looked at beat-link-trigger[0]? If the equipment you use is supported it may be suitable for your use case, it was specifically designed for coordinating effects with playback.
to be honest i was not planning to add features like that but it's very tempting. Supporting OSC is easy but a proper and useful mapping of parameters will take more brain work. i used OSC way back with PD and Reaktor.. the network layer adds latency with is ok with video sync... i have to revisit, it's years back.
thanks for your feedback, i added this topic to my backlog.
This is something that I’m interested in as well. My wife generally uses QLab for this and gets a lot of mileage out of it, but often enough ends up wanting this or that extra feature that just isn’t there.
I’m not sure that integrating it into a DAW is the way to go, but there’s definitely a need there.
I actually use Reaper a lot for something else and tried to use it for gigs too but Reaper seems to be able only to receive osc commands, couldnt find a way to send any. I think its possible to write custom cpp extension for that but cpp is not my comfort zone.
you could edit anything automatically. import -> analyse -> auto edit -> assemble -> export. this what the prototype in python used to do. the result however often needs small edits and audionaut's UI is supposed to help with that.
Some feedback after using it for a few minutes; It's a good effort and the software is useable. I understand there is a lot more to come. it would be nice to have:
- ability to import files into tracks from a menu or context menu
- keyboard short cut keys for for splitting, etc.
- automatic crossfade when moving clips into each other
- envelopes
- effects
- I think Sony Vegas had one of the most intuitive audio editing experiences when working with tracks and clips, maybe adopt what worked from it.
Task I want but have been too lazy - I have a favorite podcast, that I often fall asleep to. Certain parts with music etc. wake me up. I'd like to chop those parts out as automatically as possible. Can I do that here by e.g. giving some example edits to a few files, and asking to remove similar stuff from other files?
Sounds like something a better model armed with ffmpeg would already be able to do. Run an analysis on loudness along the track, detect when speech begins/ends (with whisper) around loud segments, and compress or cut those parts out
Actually there are models for classifying sections in audio files as speech or music, for example Silero VAD or YAMNet. I am rather confident that any decent LLM can one-shot a script that downloads the model, feeds it all the podcast episodes to generate timestamps where music is playing, and uses ffmpeg to cut out those parts. For a nice touch maybe replace each music section with a few seconds of silence, add a few second fade-out to the part preceding the silence and a few second fade-in after.
an interesting use case... the analysis could detect rhythmic or not and then edit the rhythmic parts away. i will look into, added to my backlog. thanks for your feedback!
Cool! As the other comments suggest, I could think of plenty of ways you could do this, probably much more efficiently than having a GUI, once you figured it out. But I would still like something that felt like less software engineering or data science and more "this is how I would do it manually in Audacity, do this for me automatically on a batch of files"
Concrats for finding the confidence. Looks really interesting. Put it on my list of projects to view during my down time. Thanks a ton for sharing. And congrats for getting something out the door.
Wow, thanks! I don't have time to look at it now, but it'll go high on my list. I had a quick look though the docs and I see it supports regions. Does it have labels too? Can I import/export labels and/or regions from file?
Perhaps an idea to create a docs/features.md to mention what it can do. Might even help LLMs picking up your project. Thanks again, I'm very exited!
This is a pretty neat looking platform, especially the production pipeline possibilities in it. This possibly should be front and center a bit more. Adding to my stack to try out.
it's a personal preference, i have been using JUCE for decades and it's been very reliable and stable with a well maintained forum. To be honest, i didn't even look the other way.
no roadmap... maybe i should work on a space-map ;D
given the feedback so far it might be a good idea focus on MIR and optimise audio analysis and don't fall for the audio plugin trap.
thanks for your feedback, very much appreciated!
Main difference is; this a desktop app, whereas audiomass is web.
In theory native apps have a higher ceiling for long processing tasks, since they can offload buffers on the disk and reuse - versus having to fit everything in browser memory.
the audio callback is lock free and the dsp is vectorised. i think you can expect very good performance. there is a dsp load meter implemented that displays spikes in the load. anyways... adding benchmarks is in my backlog.
thanks. it's not a DAW just yet ... i need plugins for my own music so plugin support is on my backlog. experience tells me that it's not trivial, latency compensation is tedious and dealing with countless plugins and it's formats opens a box of worms.
i love tackers but audionaut is not a tracker. it's a timeline-based multitrack editor in the Sound Designer II tradition: regions and playlists rather than patterns and pattern rows.
Hi, i finally feel confident enough to announce Audionaut. It's been 3-4 years in the making and my initial motivation was to edit my own multi-channel recordings with the old Sound Designer II workflow (create regions, drop to a playlist, export playlist, done). It got a little bigger now and the most current feature is the agent editing, which is imho very useful since you have the UI which always provides transparency.
To try the agent part with Claude Code (app and Node 18+ installed):
Edits land in the open project as one undo step each, and saving stays with you. On macOS the app is sandboxed, so keep projects in ~/Music.Downloads: https://audionaut.app/download.
any feedback is most appreciated
Congrats on the launch!
I tried to built something similar in go and wails but quickly hit roadblock with some critical features that required high time precision.
I wonder if I could use it for my use case, do you have any roadmap?
i have a backlog but no roadmap. audionaut is very precise when it comes to timing if that's what you mean or what's your use case?
My use case is pretty unique I guess and not something you built audionaut for. I do small live gigs in a local place where I sometimes need to operate lights, screen and play music at the same time and need something I can cue things to multiple software/devices through osc, websocket, tcp during music playback. I know there is already software for that but I didn't find anything I could run and quickly operate during the rehearsal before the event.
I was just wondering if you planed to add some features like markers or cue for integrations.
Have you ever looked at beat-link-trigger[0]? If the equipment you use is supported it may be suitable for your use case, it was specifically designed for coordinating effects with playback.
[0]https://github.com/Deep-Symmetry/beat-link-trigger
to be honest i was not planning to add features like that but it's very tempting. Supporting OSC is easy but a proper and useful mapping of parameters will take more brain work. i used OSC way back with PD and Reaktor.. the network layer adds latency with is ok with video sync... i have to revisit, it's years back.
thanks for your feedback, i added this topic to my backlog.
This is something that I’m interested in as well. My wife generally uses QLab for this and gets a lot of mileage out of it, but often enough ends up wanting this or that extra feature that just isn’t there.
I’m not sure that integrating it into a DAW is the way to go, but there’s definitely a need there.
You want REAPER
I actually use Reaper a lot for something else and tried to use it for gigs too but Reaper seems to be able only to receive osc commands, couldnt find a way to send any. I think its possible to write custom cpp extension for that but cpp is not my comfort zone.
Does it edit music/podcasts automatically, or still needs a human proxy? If it does automatically, how does AI know when to stop?
you could edit anything automatically. import -> analyse -> auto edit -> assemble -> export. this what the prototype in python used to do. the result however often needs small edits and audionaut's UI is supposed to help with that.
Congrats on the launch.
Some feedback after using it for a few minutes; It's a good effort and the software is useable. I understand there is a lot more to come. it would be nice to have:
- ability to import files into tracks from a menu or context menu
- keyboard short cut keys for for splitting, etc.
- automatic crossfade when moving clips into each other
- envelopes
- effects
- I think Sony Vegas had one of the most intuitive audio editing experiences when working with tracks and clips, maybe adopt what worked from it.
thanks so much for your feedback!
- context menu file import is a no brainer, added to my backlog.
- keyboard shortcut for split is command+e
- automatic crossfade when dragging usually works with the shift modifier key, added to my backlog
- envelops: this feature was requested by another user already, so top of my backlog.
- effects... yes, of course. i will not implement the in place processing but insert and send effect like it's done regular DAW. so stay tuned.
Task I want but have been too lazy - I have a favorite podcast, that I often fall asleep to. Certain parts with music etc. wake me up. I'd like to chop those parts out as automatically as possible. Can I do that here by e.g. giving some example edits to a few files, and asking to remove similar stuff from other files?
Sounds like something a better model armed with ffmpeg would already be able to do. Run an analysis on loudness along the track, detect when speech begins/ends (with whisper) around loud segments, and compress or cut those parts out
Actually there are models for classifying sections in audio files as speech or music, for example Silero VAD or YAMNet. I am rather confident that any decent LLM can one-shot a script that downloads the model, feeds it all the podcast episodes to generate timestamps where music is playing, and uses ffmpeg to cut out those parts. For a nice touch maybe replace each music section with a few seconds of silence, add a few second fade-out to the part preceding the silence and a few second fade-in after.
an interesting use case... the analysis could detect rhythmic or not and then edit the rhythmic parts away. i will look into, added to my backlog. thanks for your feedback!
Cool! As the other comments suggest, I could think of plenty of ways you could do this, probably much more efficiently than having a GUI, once you figured it out. But I would still like something that felt like less software engineering or data science and more "this is how I would do it manually in Audacity, do this for me automatically on a batch of files"
Concrats for finding the confidence. Looks really interesting. Put it on my list of projects to view during my down time. Thanks a ton for sharing. And congrats for getting something out the door.
Wow, thanks! I don't have time to look at it now, but it'll go high on my list. I had a quick look though the docs and I see it supports regions. Does it have labels too? Can I import/export labels and/or regions from file?
Perhaps an idea to create a docs/features.md to mention what it can do. Might even help LLMs picking up your project. Thanks again, I'm very exited!
Thanks! Regions yes, labels not yet.
Good idea on features.md. The manual has it all, but a one-page list is easier to find, for people and for LLMs. I'll add it.
This is a pretty neat looking platform, especially the production pipeline possibilities in it. This possibly should be front and center a bit more. Adding to my stack to try out.
thanks and agreed, the pipeline side is easy to miss.. i ll put it in a more prominent place in the README
It was a pleasant surprise, normally such pipelines are hyped up, end up being underwhelming from only being surface level.
If there was a video that could visualize the orchestration and composition that would be pretty wild.
This looks really compelling! Bookmarking to give it a try the next time I have a round of podcasts edits in my todo list.
JUCE is open source, but it has a lot of barriers for commercial products. Why did you choose JUCE over iPlug2?
it's a personal preference, i have been using JUCE for decades and it's been very reliable and stable with a well maintained forum. To be honest, i didn't even look the other way.
Got it! Do you have a roadmap? It seems to be a cool project
no roadmap... maybe i should work on a space-map ;D given the feedback so far it might be a good idea focus on MIR and optimise audio analysis and don't fall for the audio plugin trap. thanks for your feedback, very much appreciated!
How does this compare with audiomass?
Main difference is; this a desktop app, whereas audiomass is web.
In theory native apps have a higher ceiling for long processing tasks, since they can offload buffers on the disk and reuse - versus having to fit everything in browser memory.
thanks for jumping in and clarifying. audiomass is very impressive and i don't dare to do web since i am a c++ dinosaur ;)
nice to see a proper open-source multitrack editor that isn't tied to one OS. how does latency hold up with lots of tracks?
i have not benchmarked the playback yet. the performance should be industry standard an not a concern.
the dsp meter sounds useful - i'd still be curious to see track-count and buffer-size benchmarks when you get to them.
the audio callback is lock free and the dsp is vectorised. i think you can expect very good performance. there is a dsp load meter implemented that displays spikes in the load. anyways... adding benchmarks is in my backlog.
is audacity not multitrack?
you're right, audacity is multitrack too - i wasn't trying to imply this was the first one.
Yeah but it’s not very good, crashes often and has been acquired by a company pushing some online subscription
let's hope audacity and audionaut will happily co-exist and learn form each other.
is this a DAW? or primarily oriented towards podcasts/non-music. in any case, congrats!
thanks. it's not a DAW just yet ... i need plugins for my own music so plugin support is on my backlog. experience tells me that it's not trivial, latency compensation is tedious and dealing with countless plugins and it's formats opens a box of worms.
This is a mild understatement ... as in, possibly the understatement of the year.
thanks paul, i am very humbled to get a comment from you.
Is it just me but developers started creating video/audio editors after generative AI?
Most devs I see creating a video editor or best markdown editor out there with claude/codex.
flatpak please
aight, flatpak is on the list.
does not like a tracker in any way.
i love tackers but audionaut is not a tracker. it's a timeline-based multitrack editor in the Sound Designer II tradition: regions and playlists rather than patterns and pattern rows.
nice!