XiSRC - sample rate converter

XiSRC is a high quality software Sample Rate Converter to optimize your audio files for playback by adapting the sample rate and bit depth. High precision conversion algorithms ensure a perfectly preserved audio integrity.

 15.97 Add to cart

Add to cart

OVERVIEW

E

Compatibility:

  • Windows 7 – 10 / 32 & 64 Bit
  • Mac OSX Lion 10.7.3 – Sierra 10.12.x
E

Flat or Noise Shaping Dithering for 32 & 24Bit to 16Bit conversion

E

DSD to PCM Converter

E

PCM to DSD Converter (High Quality Sigma Delta Modulator)

E

Better than -200dB THD+N (Total Harmonic Distortions + Noise)

E

Peak normalization if the source file contains Inter Sample Peaks

E

64Bit High Precision Audio Engine

E

Bit Depth for Input and Output: 1, 16, 24 und 32Bit

E

Output Audio Formats: WAV, AIFF, FLAC and DSD

E

Input Audio Formats: WAV, AIFF, FLAC, ALAC, DSD and MP3

E

Supported DSD Sample Rates: DSD64, DSD128, DSD256 and DSD512

E

Supported PCM Sample Rates: 44.1, 48, 88.2, 96, 176.4, 192, 352.8 and 384 kHz

E

Metadata transfer (album, title, artist, cover, etc.) between all file formats that include metadata

E

Multi-Threading to allow the conversion of several audio files in parallel

E

Batch processing with freely configurable output file names

E

The Test Version is limited to one minute output file length

E

The activation key allows three parallel installations (Windows & MacOS X)

Apple MacOS X Image - v1.3.5

(minimum 10.7.3 – Lion)
» Release notes

Download

Microsoft Windows 32/64bit - v1.3.5

(minimum Windows 7 – 32bit)
» Release notes

Download 32bit Download 64bit

 15.97 Add to cart

Add to cart

Go to the Shop

XiSRC can be fully tested before purchase. Only the output file length is limited to 1 minute.
The activation allows 3 parallel installations. In fact, it is possible to use the Windows and MacOS X Version in parallel by just using one activation key.

WHY DO I NEED A SAMPLE RATE CONVERTER?

There are a couple of reasons why it could make sense to convert the sample rate and bit depth:
N
Use a dedicated Sample Rate to hit the sweet spot of your digital to analog converter. Each DAC has a native sample rate where it works best and therefore provides the optimal results.
N
A bit depth conversion from 24Bit to 16Bit can save disc space if the audio file has to be prepared for portable usage. The XiSRC Noise Shaping algorithm is able to preserve a high signal to noise ratio (SNR) within the most audible audio spectrum, even if the bit depth is reduced by 8Bits.
N
Some Digital to Analog Converters does not support all sample rates. In those cases the audio player or operating system needs to change the sample rate on demand. That is never an optimal process because a well-done conversion between different sample bases (e.g. 44.1kHz => 96kHz) needs a lot of processing power.
N
Most professional and home recordings are done in 24Bit/96kHz. To convert them into the digital standard format of 16Bit/44.1kHz, used for CD, MP3 or streaming, a high quality SRC is inevitable.
N
Increasing the sample rate moves the music spectrum further away from the Nyquist frequency and therefore helps to avoid the negative impact of low pass interpolation filters used to convert the signal back into the analog domain.
N
In the case of an up-sampling from a 16Bit audio file (e.g. 44.1kHz => 96kHz) it could make sense to chose 24Bit as target bit depth. By doing so the THD+N (Total Harmonic Distortions + Noise) of the target file will be reduced.

BUT WHAT IS SO SPECIAL ABOUT A SAMPLE RATE CONVERTER

Converting the sample rate within the same sample base is pretty straightforward. If we want to convert from 44.1 kHz to 88.2kHz then we just need to double the sample rate.

Well, it is not really that easy because the algorithm needs to filter the signal, after it has been stuffed with zeros, to reject images within the spectrum.

It becomes even more complex if we want to change the sample base from 96kHz to 44.1kHz. To do it in the right way we have to up-sample by a factor of 147 and afterwards we must apply a down-sampling by 320 (96kHz x 147/320 = 44.1 kHz). Such a processing involves different up- and down-sampling steps with several low pass filters involved.

XiSRC uses a 64Bit high precision audio engine to make sure that the numerical rounding errors, introduced during thousand of multiplications within the filters, are well below the signal level. The noise floor of a 64Bit number is at around -384dB that is far below the -192dB quantization noise of a 32Bit audio signal.

But now comes the next challenge. We must change the bit rate from 64Bit to 32, 24 or 16Bit. If we just truncate the excessive bits then we get something called quantization noise. This is really ugly because of the none-linear and signal correlated nature of quantization noise it creates harmonic distortions audible during faint parts of the music.

Luckily there is a mathematical trick to decorrelate the noise from the signal by just adding further noise to it before it gets truncated. That method is called “Dithering”.

There are different kinds of Dithering, but we decided to implement “Flat Dithering” and “Noise Shaping”.

Noise shaping is a feedback process where the small errors between the original and the truncated signal are fed back into the process to shape the quantization noise in a way that it decreases for important frequencies but increases for less audible frequencies (e.g. above 10kHz at a 44.1kHz sample rate).

WHY SHOULD I USE XiSRC AND HOW DOES IT WORK?

XiSRC provides great usability and quality by applying extreme precise algorithms for a very affordable price.

The key features of a THD+N better than -200dB, advanced Dithering Modes, 32Bit support and fast processing by using a 64Bit High Precision Multithreading Audio Engine are huge advantages of the XiSRC Sample Rate Converter. Additionally XiSRC uses several Threads/CPU-Cores to increase the conversion speed.

 

Six easy steps to convert your audio files:

Drop your audio files into the batch processing list or use the “Load Dialog” to select one or multiple files

Select the Target Sample Rate

Select the Target Bit Depth

We recommend to stay with the bit depth of the source material (No change) or to choose 24Bit.

Choose a Dithering Method if you do a bit depth reduction (32 & 24Bit -> 16Bit)

In most cases “Flat” is a good choice, whereas “Noise Shaping” moves a part of the quantization noise from the audible into the less audible frequencies.

Select the output file format and folder

Press “Start” to initiate the conversion of the loaded audio files

p
In-depth technical data:
There are a couple of important parameters to get an idea about the quality of a Sample Rate Converter.
N
The amount of THD+N (Total Harmonic Distortions + Noise) caused by truncation and the dithering process.
N
The phase response of the low pass interpolation and band limiting filters
N
How well is Aliasing suppressed to avoid any audio artifacts. This is defined by the stop-band attenuation.
N
The length of the FIR-Filter impulse response that could create pre-ringing because of its none causal nature.
N
The pass band for the selected sampling rate (e.g. 0 – 20kHz at 44.1kHz) as well as the pass band ripple to assure a distortion free transmission of the in-band signal.
p
XiSRC stays within the following specification:
N
The THD+N (Total Harmonic Distortions + Noise) is better than -200dB
N

XiSRC uses linear phase FIR-Filters. Therefore, the phase response is linear and does not influence the music signal.
There are a lot of controversy discussions whether the pre-ringing caused by linear phase filters is more audible than the coloration of minimum phase filters.

N
The stop-band attenuation is better than -200dB to avoid any aliasing.
N

We did everything to keep the FIR-Filters as short as possible to minimize pre-ringing. Because Sample rate Converter filters work around the Nyquist frequency the pre-ringing impacts mostly the ultrasonic frequencies, mitigating the audible effect.

The current version of the XiSRC does not provide minimum phase filters but that could be an additional feature for upcoming versions to avoid any pre-ringing at the costs of introducing coloration due to phase shifts.

N

XiSRC shows the following pass band transitions with a maximum ripple of 0.01dB:

Sample Rate [kHz] Transition [kHz]
44.1 20
48 22
88.2 40
96 44
176.4 80
192 88
352.8 160
384 176

XiSRC TESTSIGNAL MEASUREMENTS

We implemented an internal test signal generator to allow the verification of the XiSRC Sample Rate Converter at any time.

A right mouse click within the list opens a context menu. Selecting „Create Test-Signal“ opens a GUI where it is possible to choose between a sinus and sinus sweep. After setting all parameters (Sample Rate, Amplitude, Test-Signal Frequency and Target File Name) the test signal appears in the list, similar to any other audio file loaded. Finally, it is possible to convert the test signal into any sample rate or bit depth for further investigations with software like the MusicScope to verify the XiSRC performance.

Sigen Test Signal Generator
The subsequent measurements are screenshots from the MusicScope in “-200dB Mode”. The resolution allows an extreme precise analysis of the THD+N and aliasing artifacts.
1. Measurement of THD+N (Total Harmonic Distortions + Noise) by using a nearly full scale 1kHz test signal at (-0.1dB; 32Bit / 96kHz) sub-sampled to 44.1 kHz.
The MusicScope screenshot shows no harmonic distortions above -200dB. The visible distortions symmetrical to the sinus signal are caused by the FFT-Windowing of the MusicScope.
2. Sinus sweep (-0.1dB; 32Bit / 96kHz) between 0Hz and 48kHz sub-sampled to 44.1kHz to reveal any aliasing or THD+N.
The MusicScope screenshot displays the spectrogram of the sweeping sinus without any artifacts proving that there is no aliasing or any harmonic distortions above -200dB.
3. Measurement of white Noise (24 Bit / 96kHz) sub-sampled to 44.1kHz to demonstrate the pass-band.
The MusicScope screenshot shows the spectrum of the sub-sampled white noise that is rolling of at 20kHz.
4. Conversion of a 1kHz (0.1dB; 24Bit / 96kHz) sinus to 16Bit by using the different Diterhing methods “Flat” and “Noise Shaping”.
The Flat Dithering achieves an SNR of around -125dB over the whole spectrum.
The Noise Shaping moves noise from below 16kHz to the higher frequencies where it does not have such an audible effect. We measure -155dB at 500Hz and -125dB at 20kHz. The noise within the ultrasonic frequencies increases to -120dB at 48kHz.
Share This