Hello,
I converted a set of DSF tracks to 24/96 PCM (FLAC) using dbPoweramp. After conversion, I analyzed the spectrum of the resulting FLAC files and expectedly found that there are lot of undesired ultrasonic frequencies above 35kHz.
To filter these frequencies, I re-converted the DSF files, this time with applied a DSP low-pass filter set to 35kHz, which perfectly elliminated the ultrasonics from the PCM tracks. However, I found that much lower (audible) frequencies were affected as well, which was strange.
For comparison, I post-processed the unfiltered dbPoweramp FLAC file using Audacity and its Low-pass filter effect, configured to 35kHz with steep roll-off (36db per octave). The resulting spectrogram of the audible spectrum now perfectly ressembled the original file, however without undesired ultrasonics.
To illustrate the differences, I combined spectral fragments of these three scenarios into a single image:

Is this an expected behaviour from the dbPoweramp low-pass filter or I did something wrong?
Thanks!
I converted a set of DSF tracks to 24/96 PCM (FLAC) using dbPoweramp. After conversion, I analyzed the spectrum of the resulting FLAC files and expectedly found that there are lot of undesired ultrasonic frequencies above 35kHz.
To filter these frequencies, I re-converted the DSF files, this time with applied a DSP low-pass filter set to 35kHz, which perfectly elliminated the ultrasonics from the PCM tracks. However, I found that much lower (audible) frequencies were affected as well, which was strange.
For comparison, I post-processed the unfiltered dbPoweramp FLAC file using Audacity and its Low-pass filter effect, configured to 35kHz with steep roll-off (36db per octave). The resulting spectrogram of the audible spectrum now perfectly ressembled the original file, however without undesired ultrasonics.
To illustrate the differences, I combined spectral fragments of these three scenarios into a single image:
Is this an expected behaviour from the dbPoweramp low-pass filter or I did something wrong?
Thanks!
Comment