A workflow and digital filters for correcting speed and equalisation errors on digitised audio open-reel magnetic tapes - Audio Samples

  • Niccolò Pretto (Creator)
  • Nadir Dalla Pozza (Creator)
  • Alberto Padoan (Creator)
  • Anthony Chmiel (Creator)
  • Kurt James Werner (Creator)
  • Alessandra Micalizzi (Creator)
  • Emery Schubert (University of New South Wales) (Creator)
  • Antonio Rodà (Creator)
  • Simone Milani (Creator)
  • Sergio Canazza (Creator)

Dataset

Description

This repository makes available the audio samples related to the paper: Niccolò Pretto, Nadir Dalla Pozza, Alberto Padoan, Anthony Chmiel, Kurt James Werner, Alessandra Micalizzi, Emery Schubert, Antonio Rodà, Simone Milani and Sergio Canazza, A workflow and digital filters for compensating speed and equalisation errors on digitised audio open-reel magnetic tapes, Journal of the Audio Engineering Society, Special Issue on Audio Filter Design, 2022. The experiment and the three case studies are described in the publication above. This repository contains two main directories (bold indicates directory names): Experiment Samples: the 10 seconds long samples used in the experiment; Long Samples: the original 6 minutes long samples, one for each of the identified cases. Here is the notation of the file naming: W: recording (writing); R: reproducing; 3: 3.75 NAB; 7N: 7.5 NAB; 7C: 7.5 CCIR; 15C: 15 CCIR. For what concerns the Experiment Samples directory, here is the summary of the recording/reproducing standards of the adopted samples and their notation: SET A: Recording 3.75 NAB (W3) - Reproducing 7.5 CCIR (R7C); SET B: Recording 3.75 NAB (W3) - Reproducing 15 CCIR (R15C); SET C: Recording 7.5 NAB (W7N) - Reproducing 15 CCIR (R15C). Here are the variants of the samples: REFERENCE: produced by using the correct equalization standard; ANCHOR: the "Reference" altered with a low-pass filter, with pass band set at 7 kHz for music and 3.5 kHz for speech; INCORRECT: produced by using an intentionally incorrect equalization, created by mismatching the recording and reading curves and resampled to the correct speed; MATLAB: the “Incorrect” variant corrected by means of a Matlab script; API: the “Incorrect” variant corrected by means of an ad hoc web interface adopting Web Audio API, for simulating real-time correction in web applications. Here is the samples list: SET A: Training: sample4: Richard Wagner - Ride of the Valkyries; Test: sample1: Taylor Swift - Shake It Off; sample5: Queen - We Will Rock You; sample8: Bruno Maderna - Continuo; sample9: Luciano Berio - Différences. SET B (the track title reflects the name of the file from which the track itself was extracted, from the CLIPS project of the University of Napoli: http://www.clips.unina.it/en/index.jsp): Training: sample22: CLIPS project - LP4m18bZ; Test: sample15: CLIPS project - LP1f20bZ; sample16: CLIPS project - LP4m20bZ; sample17: CLIPS project - LP1f19bZ; sample18: CLIPS project - LP4m19bZ. SET C: Training: sample3: Carl Orff - Carmina Burana - O Fortuna; Test: sample2: The Weeknd - Save Your Tears; sample6: Eagles - Hotel California; sample10: Bruno Maderna - Musica su Due Dimensioni; sample12: Bruno Maderna - Syntaxis.
Date made available7 Feb 2022
PublisherZENODO

Cite this