What’s new in Oscivara 2.9

Record volume and pan movements

Double-click a track header to open its Track mixer. You can also use the sliders button in the track row. Enable Live mixing and find Automation recording.

• Read plays the saved curves.
• Touch records while you hold the control, then returns to the existing curve when you release it.
• Latch keeps recording the last value after the first touch, until you stop playback.

Start playback, move Volume or Pan, then press Stop. The completed pass is saved with the project and can be undone in one action. You can restrict recording to a range. Stop playback before saving a named version or exporting, so that the current pass is committed.

Track mixer → Automation recording. Choose the mode and range before recording a pass.

See automation recording for examples and details. The track mixer guide covers the other controls, effects and sends.

Arrange tracks in your preferred order

Drag the dotted handle beside a track name up or down and release it at the highlighted line. Audio and MIDI tracks can be interleaved; expanded automation moves with its track. The project saves this order and restores it when reopened.

For keyboard control, focus the handle with Tab: ↑/↓ move one position and Home/End move to the beginning or end. Undo/Redo also works for reordering. See the timeline guide.

Add fades to MIDI clips

Select a MIDI clip and drag the handle at its upper left or right edge. The left handle sets the fade-in; the right handle sets the fade-out. This changes the volume of that clip, including sustained notes already playing. Notes, velocities and durations remain intact.

Fades are saved with the project and included in playback and WAV export. Use the mixer or automation for changes across an entire track. Follow the clip fade instructions to get started.

Choose cloud storage and AI for your needs

Open Account → Subscription & credits. Your current plan, storage usage, balance and transaction history are together. Free offers a starting point, Cloud adds storage, and Creator and Pro combine cloud space with monthly AI credits.

One-time credit purchases remain separate from subscriptions. Current prices, allowances, conditions and checkout availability appear in that section and on the Subscription & credits page. Cards indicate when payments are unavailable. Studio and WAV export remain free.

Sign in with Google or email

Choose Continue with Google or Sign in with email in the sign-in window. Registration offers the same methods; the email form opens separately and includes password confirmation.

After a successful Google sign-in, your photo is imported if you have no avatar yet. A photo you have set yourself is preserved. If the photo cannot be loaded, sign-in still completes; import can be retried on your next Google sign-in.

Start more easily and explore loops

The empty timeline now has interactive Import audio, Add sample, MIDI and AI agent cards, plus sign-in and registration. Open Loops → Oscivara in the Studio sidebar to find material; My loops contains the upload option for your own loops.

Website loop library loading has been fixed and the Retry and Upload loops buttons improved. Previews in the library and project list now stop when you open a project or switch sections.

The Wiki includes current Russian and English screenshots and detailed guides to mix analysis, recording and export. The existing documentation structure and visual design are preserved.

Previously in Oscivara 2.8

Version 2.8 develops every music workspace around one AI agent. Requests, task progress and finished results now form the current project’s history, while audio and MIDI actions are available directly in the feed.

New Studio tools

Studio now includes live mixing, processing freeze, group buses, effect chains and gain/pan automation points. See the mixer guide for the workflow.

• Mix analysis: spectrum, live spectrogram, stereo image, peak, RMS, LUFS and mono checks. Measurements run during playback with live mixing enabled.
• Audio editor & recording: record takes, process selections, change duration and pitch, assemble crossfades and insert results into the project.
• Third-party WAM plugins: track and master effects with saved parameters and processing in the exported mix.
• Timeline: scissors in the context menu and audio clip fade handles.
• Export: 16/24-bit PCM or 32-bit float WAV, TPDF dithering, a mix preparation indicator and pre-mixer tracks in ZIP.

Aurora 2 for finished songs

SteppeMind Composer now includes Aurora 2, a new generation of song creation with expressive vocals, coherent structure and detailed arrangements. Select it in the model list and describe the instruments, mood and development. You can add your own lyrics for a song or request “no vocals” for an instrumental track.

Aurora 2 is available alongside Aurora. Preview, download or apply the result to your project; current limits and cost appear in the model panel. See the Composer guide for the full workflow.

Horizon for long music and targeted edits

SteppeMind Composer now includes Horizon. It creates long instrumental stereo tracks, continues a source recording and replaces a selected range. A negative prompt excludes unwanted traits and change strength controls how closely the result follows its source.

Aurora, Pulse, Prism and Forge remain available. Forge is suited to vocal-aware transformations and part operations; Horizon focuses on long instrumental music, continuation and targeted replacement. The selected model’s current limits and cost are shown in the interface.

Project-based AI history

The feed shows the user request, time, state and result. Concurrent tasks remain sequential cards instead of overlapping. A local project keeps its history in the browser; a cloud project keeps it with the account. After reloading, return to a result to audition, download, change or apply it.

Clearing chat requires confirmation and displays progress. Unused result files are removed with history; audio already used by the project remains protected by its project reference. Deleting a project cleans related data according to storage rules.

Results you can control

Cards use the panel width and share consistent playback, seeking and volume controls. A result can be opened for editing, separated into stems or applied to the timeline. Supported audio transformations allow model selection in the editing card.

MIDI, samples and instruments

A multi-track Score result applies to MIDI as editable tracks. Optional Maestro refinement improves instrument choice and expression and checks the resulting arrangement.

The Samples agent creates auditionable one-shots and loops. The Instruments agent creates playable, previewable instruments. Generation settings stay beside the prompt instead of appearing inside help.

Saving without empty projects

Signing in does not move an empty project to the cloud. Its first meaningful request or edit makes it eligible for saving. A non-empty local project offers cloud transfer through notifications without interrupting work.

Clear interface states

Visible states cover playback preparation, AI work, completion, errors and history clearing. Help appears automatically while a chat is empty. Result cards, top-bar actions, Library navigation and setting descriptions have been refined without changing the overall visual system.

Why Oscivara remains in public testing is explained on the Beta status page.

Continue with Models and capabilities, Composer, Maestro and Saving & storage.