Datasets:
audio audioduration (s) 4.83 123 | kind stringclasses 3
values | sensor stringclasses 9
values | description stringlengths 37 142 | source stringclasses 2
values |
|---|---|---|---|---|
solo | F3 | piano C5, pan -0.50 (front, left ear) | constructed | |
solo | F4 | piano E5, pan +0.50 (front, right ear) | constructed | |
solo | C3 | viola G4, pan -0.35 (center, left ear) | constructed | |
solo | C4 | viola C5, pan +0.35 (center, right ear) | constructed | |
solo | T3 | bassoon C3, pan -0.90 (sides, left ear) | constructed | |
solo | T4 | bassoon G3, pan +0.90 (sides, right ear) | constructed | |
solo | O1 | violin E5, pan -0.50 (back, left ear) | constructed | |
solo | O2 | violin G5, pan +0.50 (back, right ear) | constructed | |
scenario | all | alpha only under the left sensors (6 s), then only the right (6 s), then everywhere | constructed | |
scenario | all | alpha moves from the front (piano) to the center (viola), sides (bassoon) and back (violin) | constructed | |
eeg | all | simulated subject S01: 60 s eyes open, then 60 s eyes closed; the music swells when the eyes close, most of all the violins (back of the head) | mihailoba/alpha8-synthetic-eeg | |
eeg | all | simulated subject S07: 60 s eyes open, then 60 s eyes closed; the music swells when the eyes close, most of all the violins (back of the head) | mihailoba/alpha8-synthetic-eeg |
Alpha8 Spatial Sonification of Alpha EEG
Author: Mihailo Batanjac, Alpha8 team · November 2026
Most neurofeedback turns the whole brain signal into one sound. Alpha8 does it spatially: each of the eight EEG sensors plays its own instrument, placed in stereo headphones where the sensor sits on the head, and its loudness follows the alpha (8–12 Hz) power under that sensor. You hear not only how strong your alpha is, but where it changes.
This dataset is the reference design of that mapping: the table, the code, and audio examples you can listen to right here.
| Sensors | Region | Instrument | Notes (left / right) | Stereo |
|---|---|---|---|---|
| F3, F4 | front | piano | C5 / E5 | ±0.50 |
| C3, C4 | center | viola | G4 / C5 | ±0.35 |
| T3, T4 | sides | bassoon | C3 / G3 | ±0.90 |
| O1, O2 | back, where alpha is strongest | violin | E5 / G5 | ±0.50 |
Left-hemisphere sensors (odd numbers) are heard in the left ear, right-hemisphere sensors (even numbers) in the right ear; all notes belong to a C major chord, so any mix stays consonant.
Audio (headphones recommended)
| File | What you hear |
|---|---|
solo_*.mp3 |
each sensor's instrument alone, at its stereo position |
left_then_right.mp3 |
alpha only under the left sensors, then only the right, then everywhere |
front_to_back.mp3 |
alpha moves from the front (piano) to the center, sides and back (violin) |
berger_S01.mp3, berger_S07.mp3 |
60 s eyes open, then 60 s eyes closed, driven by simulated EEG from alpha8-synthetic-eeg; the music swells when the eyes close, most of all the violins |
data/metadata.csv describes every file; mapping.csv holds the sensor table with head coordinates.
How it works (sonify.py)
- Alpha power per channel in 1-s epochs (µV², 8–12 Hz).
- Gain per channel: 0 at the person's eyes-open median, 1 at four times that level, smoothed over 3 s.
- Every second each sensor plays one note of its instrument at that gain, panned with constant-power stereo panning; a light reverb ties the voices together.
Instruments are synthesized in code (no samples), the same way as in the
Alpha8 game demo.
python3 sonify.py path/to/alpha_power.csv rebuilds everything.
Limitations
The EEG-driven examples use simulated data, not recordings. The latency target of the real device (sound within about 0.1 s of the signal) is not modeled here: these files use 1-s steps for clarity. Nothing here is a medical or therapeutic claim.
License
- Downloads last month
- 21