Loading Musirion

Blog

July 18, 2026

How to Make a Piano Visualizer Video Without a MIDI Keyboard

Beautiful piano videos with falling notes, illuminated keys, colorful trails, and synchronized piano rolls have become popular across YouTube, TikTok, Instagram, and other platforms.

Creating them may appear simple, but most piano visualization software expects several separate inputs:

  • a MIDI file to determine which notes should appear and when;
  • an audio file to provide the actual sound;
  • an optional video recording of the pianist and keyboard.

This workflow is straightforward when the performance was recorded on a digital piano or MIDI keyboard. The instrument can send MIDI events directly to a computer while the audio and video are captured separately.

But what can you do if you only have an audio recording?

Perhaps the performance was played on an acoustic piano, recorded with a phone, or saved as an MP3 or WAV file. A piano visualizer cannot normally turn that recording directly into falling notes. It needs structured information about every note: its pitch, start time, duration, and hand assignment.

To create a piano visualizer video without a MIDI keyboard, the missing step is therefore to convert the piano recording into MIDI.

Converting Piano Audio or Video to MIDI with Musirion

Musirion analyzes solo piano recordings and converts them into MIDI files that can be imported into piano visualization software.

This makes it possible to create falling-notes videos from:

  • acoustic piano recordings;
  • performances recorded on a phone;
  • old MP3 or WAV files;
  • videos for which only the audio has been preserved;
  • recordings made without a MIDI-enabled instrument.

The generated MIDI is divided into separate tracks:

  • Track 0 — metadata and performance information. This includes MIDI metadata, sustain-pedal events, and other information related to the performance as a whole.
  • Track 1 — right hand.
  • Track 2 — left hand.

Separating the hands is particularly useful for visualization. Most piano visualizers can assign different colors or effects to individual tracks, allowing the right and left hands to be displayed independently.

Musirion’s current transcription quality for solo piano recordings is approximately:

  • 98% accuracy for note onsets;
  • 96% accuracy for note endings;
  • 95% accuracy when separating notes between the right and left hands;
  • 70% accuracy for sustain-pedal detection.

The generated MIDI is usually accurate enough to serve as the basis for a piano visualizer video. More difficult recordings, especially those with heavy sustain pedal, fast passages, strong room reverberation, or background noise, may still require some manual correction.

The basic workflow is:

Piano audio or video recording
        ↓
      Musirion
        ↓
MIDI with separate hand tracks
        ↓
Piano visualization software
        ↓
Falling-notes video

The MIDI does not need to replace the original sound. It controls the animation, while the original piano recording remains the soundtrack of the finished video.

Step 1: Prepare the Audio Recording

Automatic transcription works best when the recording contains only solo piano.

For better results:

  • use a clean recording with minimal background noise;
  • avoid speech, vocals, or other instruments;
  • do not cut off the beginning of the first note;
  • reduce excessive room reverberation when possible;
  • use WAV or another lossless format when available;
  • use the highest-quality MP3 available if the original recording is compressed.

Recordings made with a phone can still work, particularly when the device was placed close to the piano and the room was relatively quiet.

Step 2: Convert the Recording to MIDI

Upload the piano recording to Musirion and wait for the transcription to finish.

Once processing is complete, download the generated MIDI file. The separate right- and left-hand tracks will later make it easier to create a clear two-color visualization.

Before moving to the video stage, it is worth checking the MIDI for obvious problems:

  • missing notes;
  • extra short notes;
  • notes assigned to the wrong hand;
  • incorrect note endings;
  • pedal events that do not match the recording;
  • a beginning or ending that has been cut off.

A piano-roll editor or digital audio workstation can be used for corrections, although minor inaccuracies may not be noticeable in a simple visualization.

Step 3: Create the Visualization in SeeMusic

Once the MIDI file is ready, import it into SeeMusic.

SeeMusic is designed specifically for creating piano visualization videos. It uses MIDI data to generate falling notes and can combine the animation with audio and optional camera footage inside the same project.

Import the MIDI

Open SeeMusic and import the MIDI file downloaded from Musirion.

The note tracks should appear separately:

  • Track 1 contains the right hand;
  • Track 2 contains the left hand.

Track 0 contains metadata and pedal information rather than the notes intended for the falling-note animation.

Play the MIDI inside SeeMusic and check that:

  • the expected notes are visible;
  • the hands appear on the correct tracks;
  • the beginning and ending are complete;
  • there are no obvious extra notes;
  • the visualization remains aligned during fast passages.

SeeMusic includes MIDI editing tools, so small errors can be corrected before rendering.

Assign Colors to the Hands

Because the hands are stored on separate tracks, they can use different colors or visual styles.

A simple and readable combination might use:

  • blue or cyan for the right hand;
  • orange, pink, or purple for the left hand.

Using separate colors makes dense passages easier to follow and gives the video a more polished appearance.

Avoid using too many colors. The purpose of the visualization is to make the performance clearer, not to overwhelm it.

Step 4: Add the Original Audio

The MIDI file should control the animation, but the original recording should normally provide the soundtrack.

Import the same audio file that was converted by Musirion into the SeeMusic project.

The project will then use two synchronized sources:

  • MIDI controls the position, timing, and duration of the falling notes;
  • the original audio preserves the sound of the real instrument, room, and performance.

Align the MIDI and audio using the first clearly audible note.

If the recording begins with silence, speech, movement, or background noise, trim or offset the audio before synchronization. After alignment, mute the synthesized MIDI sound if necessary so that only the original recording is heard.

Check several points throughout the piece. If the beginning is synchronized but the end gradually drifts, one of the files may have been recorded or exported at a slightly different speed.

Step 5: Add Camera Footage

A camera recording is optional. It is possible to create a complete piano visualizer video using only MIDI and audio.

When footage of the performance is available, import it into the same SeeMusic project.

The easiest synchronization point is usually the first visible key press:

  1. locate the first note in the MIDI;
  2. find the same note in the audio waveform;
  3. align the video frame where the key begins to move.

After synchronizing the beginning, check several later sections to ensure that the hands, sound, and falling notes remain aligned.

For the cleanest result, use footage recorded from directly above the keyboard or from a slightly elevated angle. Keep the camera fixed and ensure that the full keyboard remains visible.

Step 6: Customize the Falling Notes

SeeMusic allows you to adjust the appearance of the visualization, including:

  • note colors;
  • separate styles for individual tracks;
  • two-dimensional or three-dimensional notes;
  • gradients and outlines;
  • lighting and glow;
  • particles;
  • camera movement;
  • the appearance of the virtual keyboard;
  • background colors and visual presets.

A restrained visual style usually produces the best result:

  • use one main color for each hand;
  • choose a dark or neutral background;
  • apply glow sparingly;
  • limit particles during fast or dense sections;
  • make sure individual notes remain easy to distinguish;
  • avoid effects that cover the real hands or keyboard.

If you plan to publish several videos, save the visual preset and reuse it. A consistent color scheme and layout can make a channel more recognizable.

Step 7: Export the Finished Video

Once the MIDI, audio, and optional footage are synchronized, export the final video from SeeMusic.

Recommended settings depend on the destination:

  • YouTube: 16:9, 1920×1080 or 3840×2160;
  • YouTube Shorts: 9:16, 1080×1920;
  • TikTok: 9:16, 1080×1920;
  • Instagram Reels: 9:16, 1080×1920;
  • Instagram feed: square or vertical format.

For falling-note animation, 60 frames per second generally looks smoother than 30 frames per second, particularly in fast passages.

However, 1080p at 60 fps is sufficient for most videos. Exporting in 4K is useful only when the source footage and visual elements are detailed enough to benefit from the higher resolution.

The complete process is:

Original piano audio or video
          ↓
Musirion: audio to MIDI
          ↓
SeeMusic: falling-notes visualization
          +
Original audio
          +
Optional camera footage
          ↓
Finished piano visualizer video

Limitations

Audio-to-MIDI transcription is not always perfect.

The best results are usually obtained from clean solo-piano recordings. Accuracy may decrease when the audio contains:

  • strong sustain pedal;
  • very fast repeated notes;
  • dense chords;
  • background noise;
  • heavy room reverberation;
  • other instruments or vocals;
  • clipping or compression artifacts.

Any transcription error will also appear in the visualization. An extra MIDI note becomes an extra falling block, while a missing note creates a visible gap.

For this reason, it is worth checking the generated MIDI before spending time on visual effects and final rendering.

Pedal detection is currently less accurate than note onset, ending, and hand-separation detection. This usually has a smaller effect on the falling-note animation itself, but it may influence synthesized playback if the MIDI sound is used instead of the original recording.

Conclusion

Most piano visualization tools require MIDI because audio alone does not contain an explicit list of notes, timings, durations, and hand assignments.

Traditionally, this meant that falling-notes videos had to be planned in advance and recorded using a digital piano or MIDI keyboard.

With Musirion, an existing solo piano audio or video recording can be converted into a structured MIDI file with separate tracks for the right and left hands. That MIDI can then be imported into SeeMusic to control the visualization, while the original audio remains the soundtrack.

The result is a practical workflow for turning an ordinary acoustic-piano recording into a polished falling-notes video without needing a MIDI keyboard.