BEGIN:VCALENDAR VERSION:2.0 PRODID:-//linuxsoftware.nz//NONSGML Joyous v1.4//EN BEGIN:VEVENT SUMMARY:Developing tools for the recovery of historical audio recordings DTSTART:20251118T140000Z DTEND:20251118T160000Z DTSTAMP:20260811T085007Z UID:577cd3b1-8822-4cc2-b22e-1587c7b9793a SEQUENCE:2 CREATED:20251118T090508Z DESCRIPTION:LinkMagnetic tape recordings hold historically irreplaceable a udio\, yet many recordings are unplayable due to physical and chemical deg radation\, which pose challenges to their preservation. This thesis develo ps tools for recovering audio from data obtained using non-contact readout techniques\, such as X-ray Magnetic Circular Dichroism (XMCD) and Tunneli ng Magnetoresistance (TMR) sensors. Signal denoising was tackled using thr ee methods: Extended and Unscented Kalman Filters for model-based denoisin g\, with the UKF producing cleaner spectrograms than the EKF\, but both re quiring extensive tuning\; Spectrogram soft-masking with Griffin-Lim phase reconstruction\, which effectively removed broadband noise for isolated t ones but introduced artifacts for complex sweeps\; and a two-stage U-Net ( Moliner et al.) pretrained on gramophone recordings\, which revealed obscu red harmonic content in a Beethoven excerpt\, illustrating the promise of deep learning for this task\, but still needs further validation and fine- tuning. A recording-to-readout simulation pipeline was implemented. It inc orporates the Karlqvist field\, AC bias\, hysteresis behavior via Jiles-At herton and Preisach models\, includes Mallinson-type noise\, and represent s the XMCD beam as a skewed Gaussian convolution. Experimental hysteresis loops were fitted using both models\, with the Preisach providing superior accuracy (NRMSE 3.83%) compared to JilesAtherton. This framework enables the study of particle statistics and measurement parameters on the recover ed signal. These tools form the basis for future reconstruction of real hi storical audio from degraded tapes using XMCD measurements. The simulation tools were validated on experimental XMCD scans of 1 kHz\, 1-20 kHz and m usical excerpts\, confirming the feasibility of non-contact audio recovery LAST-MODIFIED:20251118T090526Z LOCATION:Online URL:http://df.vps.tecnico.ulisboa.pt/pt/eventos/developing-tools-for-the-r ecovery-of-historical-audio-recordings/ X-ALT-DESC;FMTTYPE=text/html:

Link

Magnetic tape recording s hold historically irreplaceable audio\, yet many recordings are unplayab le due to physical and chemical degradation\, which pose challenges to the ir preservation. This thesis develops tools for recovering audio from data obtained using non-contact readout techniques\, such as X-ray Magnetic Ci rcular Dichroism (XMCD) and Tunneling Magnetoresistance (TMR) sensors.

Signal denoising was tackled using three methods: Extended and Unsc ented Kalman Filters for model-based denoising\, with the UKF producing cl eaner spectrograms than the EKF\, but both requiring extensive tuning\; Sp ectrogram soft-masking with Griffin-Lim phase reconstruction\, which effec tively removed broadband noise for isolated tones but introduced artifacts for complex sweeps\; and a two-stage U-Net (Moliner et al.) pretrained on gramophone recordings\, which revealed obscured harmonic content in a Bee thoven excerpt\, illustrating the promise of deep learning for this task\, but still needs further validation and fine-tuning.

A recording -to-readout simulation pipeline was implemented. It incorporates the Karlq vist field\, AC bias\, hysteresis behavior via Jiles-Atherton and Preisach models\, includes Mallinson-type noise\, and represents the XMCD beam as a skewed Gaussian convolution. Experimental hysteresis loops were fitted u sing both models\, with the Preisach providing superior accuracy (NRMSE 3. 83%) compared to JilesAtherton. This framework enables the study of partic le statistics and measurement parameters on the recovered signal. These to ols form the basis for future reconstruction of real historical audio from degraded tapes using XMCD measurements. The simulation tools were validat ed on experimental XMCD scans of 1 kHz\, 1-20 kHz and musical excerpts\, c onfirming the feasibility of non-contact audio recovery

END:VEVENT END:VCALENDAR