◆ Playground · real FQL encoder, in your browser

Drop a sound.
Hear it lose layers

No simulation. Your file is Haar-DWT'd into 12 bit-plane layers right here in the browser. Drag the slider to truncate. The reconstruction plays back live — and the SNR is computed against the original sample-by-sample.

▶ Try a 5-second sine sweep · 🎤 Record 5 s from microphone · View the SDK source
Source
—
Duration · SR
—s
Original size
—KB
FQL size · all layers
—KB
◆ Layers kept · drag to truncate 12 / 12
1 · coarsest · ~6 dB SNR 6 · usable 12 · full · ~70 dB
Originalreference
FQL · reconstructed— dB SNR · — KB
Same thing, via curl 
curl -X POST https://stream.mirrorfractal.com/api/encode \
  -H "Content-Type: application/octet-stream" \
  -H "X-FQL-Bits: 12" \
  --data-binary @your-audio.wav \
  -o stream.fql

# truncate to 8 of 12 layers when serving
curl -X POST 'https://stream.mirrorfractal.com/api/decode?layers=8&format=wav' \
  --data-binary @stream.fql -o reconstructed.wav
⬇ Download stream.fql · ↺ Try another file