Author Topic: Enable/disable switch for inter sample peaks and frequency respond  (Read 1684 times)

Offline geozem

  • Newbie
  • *
  • Posts: 36
Enable/disable switch for inter sample peaks and frequency respond
« on: September 19, 2015, 04:47:07 AM »
Dear Stephan

I am wondering if somebody else is giving this amount of feedback.

I had the opportunity to test different resolution convertion file results of this tool in a huge eletronic shop with different DAC from Sony, Marantz, Denon, Yamaha, Teac and Pioneer with the butifull sounding B&W Nautilus. What a fun! Thank you Bic Camera.

I found out that the DAC from Marantz, Yamaha and Sony are doing somehow different inside like the others.

The result was using the XiSRC was in case of upsammpling better but in downsampling has changed the sound very strongly. It bekame that many high frequency tones became very present/detailled/strong where the bass nearly was eliminated/faded away.
The sound bekame very glassy and very puntchy and the softness of the Contrabass gone against the orignal files.

I have used mainly Jazz recordings from ECM (original 24bit 96kHz). Upsample to 192kHz and downsampled to 44.1kHz.

After using the Yamaha DAC which has the possibility to enlarge the high freqencies (feature to compensate the mp3/aac conversion) by a button - the sound bekame warm again as the original was.

Please check under the rider "Features"

based on this experience I kindly request you to allow
Enable/disable switch individually for:
  • inter sample peaks correction and
  • frequency respond filtering

I think that it would be great feature to be enabled/disabled based on own equipment and not fixed.
This would be so far my final feedback on the features

Best Regards
« Last Edit: September 19, 2015, 05:12:54 AM by geozem »

Offline Xi_Stephan

  • Global Moderator
  • Sr. Member
  • *****
  • Posts: 455
Re: Enable/disable switch for inter sample peaks and frequency respond
« Reply #1 on: September 28, 2015, 08:35:52 AM »
Hello geozem,

Thanks for all your valuable feedback.

A sample rate converter (SRC) needs to stick to the rules of signal theory, especially the Shannon theorem (

That implies the following filtering and Inter Sample Peak corrections:


a.) The signal gets up-sampled by adding zeros between the samples. For example if you up-sample by a factor of 2 then just one zero sample will be added between any original sample.

b.) To get the final up-sampled signal a low pass filter at the original 1/2x sample rate must be applied. In the case of 44.1 kHz => 88.2 kHz we must apply a filter that removes or at least strongly attenuates all frequencies above 22.05kHz to avoid any aliasing.
The quality of the sample rate converter strongly depends on those filters.

c.) If we up-sample a signal that includes Inter Sample Peaks then those peaks get immanent and we have to remove them at the end of the process.


a.) The down-sampling process is just the reverse approach. First we must apply a low pass filter to remove all frequencies above 1/2x target sample rate and then we through away samples. In the case of a down-sampling from 88.2kHz to 44.1kHz we have to apply a filter which cuts off all frequencies above 22.05kHz and then we are allowed to remove any second sample to get the target sample rate of 44.1 kHz.

b.) Inter Sample Peaks are not an issue when we down-sample.

So, you see that there is no way to enable/disable the low pass anti aliasing filters or the Inter Sample Peak removable because it is strongly necessary to mathematically adhere to the theory of discrete signals.

Algorithms to compensate for the loss of high frequencies caused by MP3 compression are doomed to fail because MP3 and AAC thin out frequencies above a defined cut-off frequency, dependent on the chosen bit rate. There is actually no  way to enhance the sound quality by just amplifying those higher frequencies.


« Last Edit: September 28, 2015, 08:46:01 AM by Xi_Stephan »