Recording voice notes on a mix

Some feedback doesn't compress into a label. "The snare feels late and small" is three seconds of speech and a full minute of typing, and by the time you've finished typing you've lost the thing you were trying to describe. A lot of mixing notes are like that - observations about feel, about how a section lands, about what a vocal is doing emotionally. Words don't come to you in written form. They come to you in spoken form.

Voice notes in Mix Notes are just custom annotations that happen to be audio instead of text. Same timestamps, same timeline, same share behaviour. They exist because the fastest way to capture "this chorus doesn't build right" is to say it out loud while it's happening, not to pause and hunt for the words. This guide covers the recording flow, when voice beats text and when it doesn't, and how voice notes behave when you share the mix.

Recording a voice note

Voice notes live inside the Custom annotation (the "D" key, or the Custom button). Tapping it opens a sheet with a text box and a red record button. Tap the record button and the browser asks for microphone permission the first time. Grant it once and it's remembered for that origin.

Playback pauses automatically when you start recording - you'll be talking, not critical-listening, so this keeps the voice note free of mix audio bleeding into it. A timer shows seconds elapsed. The timestamp is captured at the moment you tapped record, not when you finish speaking, so the note is anchored to where in the track the problem was.

Tap the button again to stop. The voice clip attaches to that timestamp in the timeline, and playback resumes from where you paused.

Recordings cap at 60 seconds. If you have more to say than fits in a minute, that's a sign it should be two separate notes at two separate timestamps, or a written comment you can expand on in a follow-up. There's no editing or trimming inside the app - if you stumble over a word, delete the note and re-record. It's not a podcast, it's a sticky note.

See how to annotate a mix for the full capture flow and the other five annotation buttons.

When voice wins vs when text wins

The quick version:

Voice wins for:

• Emotional feedback ("this doesn't hit the way it should")
• Nuance that's hard to write ("it's almost right, but there's something in the low mids")
• Medium-length comments ("the whole bridge needs to come down in energy before the last chorus hits")
• Describing a sound you want ("more like a wet plate, shorter tail")
• Comments to a bandmate who'll understand tone and inflection

Text wins for:

• Quick markers (use the built-in labels instead: Too Loud, Tuning, Glitch, Timing)
• Precise references ("cut 3 dB at 2.4 kHz")
• Notes you'll scan through quickly later
• Anything a search should find (you can't search voice)
• Technical fixes you already know the answer to

Rough rule: if you'd naturally say it out loud to yourself, record it. If you'd naturally type it, type it. Don't force everything into one format.

The other factor is who the note is for. If the note is for you, future-you will probably read text faster than listen to voice. If the note is for a client or bandmate, voice conveys tone that text can't, and often lands better because the recipient hears you thinking instead of reading your conclusions.

Playback during review

Voice notes sit in the timeline alongside text notes. Each row shows its timestamp, and voice notes have a play button next to the delete action. Click the play button to hear the clip inline.

Mix Notes pauses the main transport while a voice note plays, then resumes when the clip ends. Trying to listen to voice commentary over the mix at the same time is confusing, and most voice notes are short enough that the pause is barely noticeable. Click a row to jump the playhead to its timestamp (offset two seconds earlier for run-up), same as any other note.

There's no filter to show only voice notes - they live interleaved with text notes sorted by timestamp. If you've captured a messy mix of both, skim down the list looking for the play icons.

Sharing voice notes

When you generate a share link, any voice notes attached to that file are available to the link recipient just like text notes.

The share page shows live notes from the owner's library. If you add a voice note in your library, the share page reflects it on the next load. If a collaborator signs in, the file is added to their library, and any notes they add (voice or text) live in their own library - see sharing mixes with clients for the full share flow.

Privacy and storage

Voice notes are stored like any other annotation. Locally they live in your browser's IndexedDB alongside the audio file blob. When you're signed in, the app uploads voice recordings to Supabase storage alongside your other annotations, so they're playable on your other devices.

Voice notes on a shared file travel to the share bucket with the audio. Deleting a voice note removes it from your library locally and drops it from cloud sync on the next annotation write. There's no revoke-share action in the UI right now, so a voice note captured and shared is effectively out there for anyone with the link until you delete the source file.

If you're working on something under NDA, keep the file local (don't share it) and your voice notes stay on-device.

What's next

Voice notes pair well with sharing mixes with clients - clients leave better feedback in voice than in text, and getting timestamped voice comments back is a noticeable upgrade on email paragraphs. Using AB mode for reference tracks is the next stop if you want to compare your mix against a reference while you're capturing observations.

Or open the app and leave your first voice note on a mix in progress.