Published January 6, 2026 | Version 0.4

MAPS - MIDI Aligned Piano Sounds

  • 1. ROR icon Aix-Marseille Université
  • 2. ROR icon Institut de Recherche en Informatique et Systèmes Aléatoires
  • 3. Télécom ParisTech

Description

MAPS, standing for MIDI Aligned Piano Sounds, is a piano sound database dedicated to research on multi-F0 estimation and automatic transcription:

  • MAPS is composed of about 31 GB of CD-quality recordings in .wav format.
  • High quality audio is obtained by means of Virtual Piano softwares and a Yamaha Disklavier.
  • Nine settings of different pianos and recording conditions were used.
  • Four categories of sounds are provided:
    • isolated notes and monophonic sounds
    • random chords
    • usual chords
    • pieces of music
  • The ground truth is provided for all sounds, in MIDI and text formats. The audio was generated from the ground truth in order to ensure the accuracy of the annotation.
  • MAPS is freely released with a Creative Commons license.

License: Creative Commons (http://creativecommons.org/licenses/by-nc-sa/2.0/fr/deed.en_US)

Author: Valentin Emiya (Telecom ParisTech, 2008) 

References:

  • Multipitch estimation of piano sounds using a new probabilistic spectral smoothness principle, V. Emiya, R. Badeau, B. David, IEEE Transactions on Audio, Speech and Language Processing, 2010.
  • Transcription automatique de la musique de piano, V. Emiya, PhD Thesis, Telecom ParisTech, 2008.

Files

AkPnBcht.zip

Files (18.6 GB)

Name Size
md5:794d58f8ea2864edad365145e6d7c637
1.7 GB Preview Download
md5:8a2147f9a146da7bef05accf47015d90
1.8 GB Preview Download
md5:c1ec6d005688c16dbddaccca03d30066
1.8 GB Preview Download
md5:d0f63568621bce3a48e175dbff90e680
2.1 GB Preview Download
md5:a1c8c9945a2c605f62484b80cfa20cf4
2.7 GB Preview Download
md5:72bbdf40eb7af69225755e165a0a0a08
2.6 GB Preview Download
md5:caeea03000812bbc1e82cbeff1d8c102
652 Bytes Download
md5:ca887f61e0ed88965cf1e31feafe99e3
18.6 kB Preview Download
md5:74e1fdc9937de7ab149bbc6d139f9b4b
1.7 kB Preview Download
md5:eaeb055a5e26ca4b1635036999c6e49e
1.8 GB Preview Download
md5:7cfdfe0bd523307e9ac53dde890396e8
2.1 GB Preview Download
md5:33adeee52a04572379167db91e8b13de
2.0 GB Preview Download

Additional details

References

  • Multipitch estimation of piano sounds using a new probabilistic spectral smoothness principle, V. Emiya, R. Badeau, B. David, IEEE Transactions on Audio, Speech and Language Processing, 2010.
  • Transcription automatique de la musique de piano, V. Emiya, PhD Thesis, Telecom ParisTech, 2008.