AudioDistillery v0.9.0-rc.2

Second release candidate: everything from rc.1 (full-pane Record surface with live monitor meters, mid-take lane renaming, import zone, startup and hot-plug fixes) plus two track-list fixes — double-click renaming no longer touches playback, and clicking a title scrolls the waveform to the playhead (zoom preserved).

AudioDistillery v0.9.0-rc.1

Pre-release for testing the new full-pane Record surface (live monitor meters, mid-take lane renaming, import zone), the startup-flash fix and hot-plug under monitoring. Not production — feedback welcome.

AudioDistillery v0.8.0

The editing release: AudioDistillery now shapes the album, not just the tracks.

Clip editing

Select any clip — its frame, its name badge, or the new ⋯ menu at its top-right — and delete it: the timeline ripples closed across every layer (aligned takes disappear together), markers follow, and one ⌘Z restores everything instantly, audio included. Or turn the clip into a titled track in one click. Importing several files gains "One after another — as album tracks": one clip and one titled track per file — an album session in a single gesture.

Selection that behaves

Selections butt against existing tracks — no gesture can draw an overlap. Edge drags snap magnetically to clip boundaries (the whole frontier lights up), dragging near the viewport edge auto-scrolls, Shift-click completes a selection from your M-mark, the playhead or the nearest selection edge, and the numbered edge flags are now grab handles — no more hunting a one-pixel line.

The album, for real

• Gaps between titles: a per-album default (2.0 s) with per-boundary overrides — segues welcome — edited in the track list, the album strip or the Album dialog. CD images and DDP filesets press them as real Red Book pregaps (INDEX 00 in cue and PQ subcode — players count them down, rippers show them). Per-track files stay byte-identical, gaps or not.
• The target timeline: an album strip shows the record as it will be delivered — titles, gaps, disc breaks — updating live with every edit.
• One player, two programs: an explicit Source | Album toggle. Album mode plays the delivered sequence — mid-file track starts, real silences — with ⏮ ⏯ ⏭ track navigation in both modes. Seeks never change the program behind your back.

Workflow phases

A Record · Edit · Master switcher (keys 1/2/3) that emphasizes and never walls off. Edit collapses mastering to an always-visible meter rail; Master enlarges the album strip, surfaces gaps and ISRCs in the track list, and turns Export into the phase's call-to-action with an album-readiness checklist (badge on the button, details on hover, every open item clickable to its fix). Each phase picks its default program; your explicit choices persist. The waveform now separates the clip identity band (top) from the track windows (below) — the two layers never blend again.

Recording, everywhere

• Windows (#9): ASIO backend for true multichannel interfaces (WASAPI stays the fallback), multi-host input listing, and event-driven device hot-plug — verified in a Windows 11 VM.
• macOS fix: recording captured silence because the app never declared microphone access. The permission is now properly declared and requested, the record dialog shows the real authorization state with "Grant access" / "Ask again", and a take warns when every lane stays silent.

"Negative feedbacks" always welcome ;)

AudioDistillery v0.7.0

AudioDistillery now starts where the rehearsal starts: it's the recorder too.

Multitrack recording (#8)

Press ● Rec (toolbar, or the empty-session screen) and track your rehearsal like a tape machine:

• Pick your interface, build your lanes. Inputs are listed by their real driver names on macOS ("MIC/LINE 1", "S/PDIF L", "VIRTUAL 3", …). Add a linear batch of lanes from any starting input, then repoint or rename any lane individually — non-contiguous mappings are fine. Lane names become layer names, and your setup is remembered for the next session.
• Reliability first. One sample-synced mono 24-bit WAV per lane at the device's native rate, streamed to a fresh "Take N" folder. File headers are fixed up every couple of seconds, so a crash or a full disk never loses the take beyond its last flush. Existing files are never touched.
• See what you're tracking: per-lane peak meters (red above −3 dBFS), elapsed time, and a dropped-frame counter that should stay at zero.
• Hot-plug aware: plugging or unplugging an interface updates the device list by itself — event-driven, no restart, no polling in the UI.
• When you stop, the take opens directly as a synced multitrack session on the per-layer view — ready for auto-split, per-track mixing, mastering and export.

Input monitoring and record-path plugins are deliberately out of scope for now: nothing is allowed to compete with the recording.

Also in this release: CI fixes (Windows path assertion, resilient Linux ffmpeg download).

"Negative feedbacks" always welcome ;)

AudioDistillery v0.6.0

One feature, but a big one: AudioDistillery now produces the file a pressing plant actually asks for.

DDP delivery (#5 — issue complete)

Pick "DDP fileset + PQ sheet" in the export dialog (WAV · 44.1 kHz · 16-bit) and you get a full DDP 2.00 fileset:

• IMAGE.DAT — the audio image in 2352-byte CD sectors, initial 2-second pause included, dithered from the full mastering chain like every export;
• DDPID, DDPMS, PQDESCR — the descriptors and PQ subcode stream, with track positions exact to the CD frame;
• CD-Text — album and track titles, performer, EAN and ISRCs, visible in any DDP player;
• CHECKSUM.MD5 — verifiable with md5sum -c;
• PQ_SHEET.TXT — the human-readable companion: every track's start, length, ISRC, plus the EAN — what you print and sign off with the plant.

It's validated the paranoid way: the integration suite re-reads the generated image and null-compares it sample-by-sample against the WAV produced by the cue path from the same session, and checks that the PQ stream agrees with the map to the frame. Filesets open cleanly in third-party DDP readers (XLD shows tracks, durations and CD-Text).

Per-track ISRC

A new editor in the export dialog (visible for CD deliverables): ISRCs are validated (CC-XXX-YY-NNNNN), normalized, saved in the project, and embedded in the DDP subcode, the PQ sheet and cue sheets as ISRC lines.

Under the hood

The DDP writer is a standalone MIT crate, ddp-fileset — no dependency on the app, reusable by any Rust project that needs to master CDs.

"Negative feedbacks" always welcome ;)

AudioDistillery v0.5.0

This release turns the export side of AudioDistillery into a proper delivery tool: professional formats, real metering, and multitrack stems.

More export options (#5, tiers 1–2)

• Sample-rate conversion on output (large-kernel aresample) and dither when reducing to 16-bit — auto, triangular, triangular HP or Shibata.
• A one-click CD preset (WAV · 44.1 kHz · 16-bit · dithered), and a full Red Book CD image + cue sheet export: one frame-aligned WAV (588-sample boundaries, engine-side SRC so the cue offsets are exact), CD-Text, and your album's EAN/UPC as CATALOG.

Metering (#2)

• A live EBU R128 meter in the Chains panel: momentary and short-term bars, integrated LUFS with a calibratable target (Spotify, Apple Music, YouTube, EBU R128 presets), loudness range, and max-hold true peak — tapped right after the master bus, so it reads exactly what your export will contain.
• Every export now ends with a loudness report: LUFS-I and true peak of each delivered file (measured after encoding — codec overshoots included), album outliers flagged, anything above −1 dBTP in red. CD images are measured per cue segment, and a visible "analyzing" phase replaces the old silent pause before the report.

Multitrack stems export (#7)

• Export your cut tracks as stems: one file per layer, one folder per track (01 - Song/01 - Room mic.wav), with {layer} and {ln} macros for custom naming.
• Stem content is your choice: raw cuts (sample-exact slices of the source files, ideal for re-mixing in a DAW) or with mix settings (layer gain/mute/solo and layer inserts). A layer with no audio under a track still yields a full-length silent stem, so the set always re-imports time-aligned.
• A new "Source" format choice mirrors each original file's container, bit depth and sample rate — per layer in stems mode.

Also fixed: the load overlay's Cancel button could swallow a stray double-click right after choosing a layout and abort the analysis; it now arms after a short delay.

"Negative feedbacks" always welcome ;)

AudioDistillery v0.4.0

AudioDistillery 0.4.0 — faster ends of rehearsals

Two features aimed squarely at the core workflow: turning a long recording into named songs, quickly.

Renameable layers (#3)

Double-click a layer's name to call it "Room mic" or "Bass DI" instead of recorder-input-3.wav — same Enter/Esc/Tab flow as track renaming (Tab hops to the next layer), original file name kept as a tooltip, and the name follows everywhere: per-track mix rows and the Chains panel.

Auto-split, evolved (#4)

• Pick the detection source in multitrack sessions: the mix, or any single layer — a between-songs-quiet input (a DI, a spot mic) often detects songs a bleeding mix can't. The bar opens even when nothing is found, so switching the source is one click.
• Review the proposals by ear before adding: navigation follows the playhead (‹ › jump between proposals, ⇥ auditions an ending, the indicator reads –/N when you're outside every proposal), and each proposal can be kept or excluded — excluded spans stay visible in red on the waveform, and "Add N tracks" adds exactly what you kept.

Includes the 0.3.1 Windows playback fix for anyone jumping versions.

"Negative feedbacks" always welcome ;)

AudioDistillery v0.3.1

AudioDistillery 0.3.1 — Windows can hear again

Patch release fixing one real bug found by a user report:

Fixed

• No playback audio on Windows (#6): the engine forced the session's sample rate onto the output device, which WASAPI shared mode rejects (macOS never showed it — CoreAudio was silently resampling for us). The stream now opens at the device's own rate, with a streaming windowed-sinc resampler converting session → device on the render thread. Identical, explicit behaviour on every platform, verified inside a Windows VM by probe and by ear.
• Silent failures are gone: if the output device refuses the stream (or disappears), the status bar now shows an actionable "Audio device error" message instead of mute silence.

Exports are unaffected — they never go through the output device.

"Negative feedbacks" always welcome ;)

AudioDistillery v0.3.0

AudioDistillery 0.3.0 — the mastering chains era

First release since AudioDistillery went open source (MIT) — and a big one. Everything that happened since v0.1.0:

Real-time mastering with your own plugins

• Audio Units and VST3 hosting, with the plugins' native editors — tweak while you listen.
• Insert chains at three scopes: per layer (pre-fader), per track (active only inside that track), and on the master bus — managed from the new Chains panel (Master / Layers / Tracks, the Tracks section following the playhead).
• Chain presets: save a chain with its live plugin states, reload it on any target in any project.
• Chain latency display; exports render through the exact same signal path, latency-compensated — null-tested bit-perfect between the AU and VST3 builds of the same plugin.
• Custom VST3 scan folders, and a plugin picker that merges both formats per manufacturer with AU/VST3 badges.

Album finishing

• Format-agnostic tagging (ID3v2 / MP4 / Vorbis / RIFF INFO), multi-disc numbering, filename macros, cover art — written correctly, checked against the pickiest consumer (Apple Music).

App & project

• In-app About (version, licenses); click the Still wordmark.
• Clicking a track starts playback when the transport is stopped.
• Projects migrate automatically (v7); your source files remain untouched — as always, enforced by the test suite.

Open source

• MIT license, ARCHITECTURE.md, CONTRIBUTING.md, CHANGELOG.md — and the story of why this exists.

macOS builds are signed and notarized. On first launch of the unsigned Linux/Windows builds, the usual platform warnings apply.

"Negative feedbacks" always welcome ;)

AudioDistillery v0.1.0

v0.1.0 — first public build (splitting, multitrack mixing, metadata, …