Jump to content
Computer Audiophile


  • Content count

  • Joined

  • Last visited

1 Follower

About audiventory

  • Rank
    Yuri Korzunov, software developer

Recent Profile Visitors

The recent visitors block is disabled and is not being shown to other users.

  1. audiventory

    16 bit files almost unlistenable now...

    By available publications, editing 1 bit DSD to 1 bit DSD without re-modulation here is cut/merge only. Mixing 1 bit DSD to 1 bit DSD is impossible without re-modulation. Without filtering, part of dynamic range is lost. Filtering give ringing. However, mixing (both PCM and DSD) can use digital filters and have ringing and amplitude/phase distortions. It may become matter comparing with filtering DSD. Analog mixing have non-linear and frequency distortions, noise. When record is played back PCM converted to DSD with ringing during upsampling. Pure PCM R2R DAC have more issues, than DSD modulated. Resume: If record is edited, I don't see problem in accurate conversion DSD to PCM for editing comparing other issues in given context.
  2. audiventory

    16 bit files almost unlistenable now...

    Looks like I done something wrong when quote.
  3. audiventory

    16 bit files almost unlistenable now...

    To estimate quality, it is necessary to define measurable value (distortions, method of audio perception testing, etc.).
  4. audiventory

    16 bit files almost unlistenable now...

    Noise shaping is defined by filter in feedback of a sigma-delta modulator. See "DAC" in feedback at the picture "Sigma delta modulator inside" here https://samplerateconverter.com/educational/how-work-sigma-delta-modulation-audio Remark: there are number of different modulator schemes, but in the post described general principle. More shaping steepness (filter frequency-magnitude steepness) > more sensitive feedback > more unstable system with feedback (modulator). Steepness of the shaping depend on the order and design method of the filter. 1 order = 1 [multiply+sum] operation. As rule, sigma-delta modulators have orders lesser 10. And changing of the shaping steepness don't make big performance difference. Resampling filters and big sample rates (big number of calculations) consume most of performance.
  5. audiventory

    16 bit files almost unlistenable now...

    ADC distortions also depend on higher sample rate. For some kinds of editing sample rate is not big matter though.
  6. audiventory

    16 bit files almost unlistenable now...

    In high resolution, It is easier to implement lesser distortions technically. For me 44.1 kHz and DSD64 are most sophisticated in implementration.
  7. audiventory

    16 bit files almost unlistenable now...

    Dynamic range is not definded as difference between noise floor and peak level. It definded difference between levels with allowable distortions. Read details here https://samplerateconverter.com/educational/dynamic-range I think, when acoustical noise level (air pressure) of record on 50 dB, it's too much, even worse than analog sources. I suppose, 50 dB of the acoustical pressure is like to cooling fans noise at hard working notebook. Probably, noise floor may be from 0...10 dB in the listening room.
  8. audiventory

    DSF or DFF. What is the difference?

    Check my software https://samplerateconverter.com/dsf/dsf-dff-converter For exact binary audio content in source and destination files, set bit-perfect mode (in commercial versions only) https://samplerateconverter.com/iso-converter/convert-iso-dsf-wav-flac-aiff#bit-perfect-iso-conversion
  9. audiventory

    The advantages of a NAS?

    First, I'd pay main attention to electrical energy consuming of NAS. It is online permanently.
  10. It's code C that converded to gain coefficient to multiply with input signal X. Y=G(C)*X.
  11. If electrical dynamic range of DAC differ dynamic range of next amp, analog gain control is need there. If these dynamica ranges are fitted, no difference analog or digital gain control there.
  12. dB volume control allow to alter volume in scale that close to human ear. It don't differ from linear gain control by quality or other features. To simpler understanding, you can consider bit depth just as noise level. Sine in analog and digital form is Y=sin(X). Gain control in analog and digital form is Y=sin(X)*G. Gain control (dB) in analog and digital form is Y=sin(X)*10^(GdB/20). X - input Y - output G - gain linear GdB - gain in dB If bit depth is accounted (Qn - quantization noise) Y=sin(X)+Qn(sin(X)) Y=sin(X)*G+Qn(sin(X)*G) Y=sin(X)*10^(GdB/20)+Qn(sin(X)*10^(GdB/20)) If these operations are performed in float point formats (in programming meaning) and Y is rounded correctly, these operations are similar to analog volume control (accounting quantization). If these operations are performed in integer formats (in programming meaning), there is used wider bit depth for calculation and rounding, when bit depth is truncated. Precision depend on how many bits are used for calculations. It is matter or used bit number, to works in integer like float point. The bit number (calculations/source/target) define Qn. Resume: Result of gain processing of input signal and quantization noise are separate. And signal should be considered as uninterrupted in time and level. Bit depth of source/processing/target impact to quantization noise level. Quantization noise in uninterrupted too and added to uninterrupted signal (see goal 1).
  13. Two signal dynamic add information, but it can't say about sound quality. It is only compatibilty to overload. Sound quality is defined non-linearity in frequency range from 0 to infinity. And non-linearity have different types that may not be finally compared pure mathematically. I'd want to correct characteristics that I noted above: to estimate sound quality of device, it is need input/output amplitude responces at all frequencied. It cover all device compatibilities. But it can't finaly estimate device mathematically again.
  14. Digital volume control may be implemented without difference for 16 and 24 bit. Because digitally coded analog signal is uninterrupted in time and level mathematically. But there are implementations that makes difference is possible. Dynamic range is defined by bit resolution. If digital volume control is implemented properly, dynamic range is defined by audio resolution solely.
  15. a) To get more information apparatus anslyzed in frequency-amplitude/phase-time domain. b) There is need analyze wave field. I'm not sure that it is possible in current technical level. c) In general case, reference tool must have precision 3-10 times more than expected deviation of measured value. If you want measure with 10 dB precision,measurement tool must provide 1-3 dB error. To comparison of harmonics, it is need to get reference level in V (Volt) or dBV (dB 1 Volt). Because at different levels, spectral content is various. d) Sweep sine may be use to auto dynamic range estimation. Manual analyzis is too complicated. Most full feature of apparatus are commbination of minimal and maximal values at each frequency. Minimal value may be accepted as value that provide N dB signal/noise ratio. Maximal value may be accepted as value that cause M dB harmonics above noise. Read more about dunamic range https://samplerateconverter.com/educational/dynamic-range Also measured values should be calibrated with psychoacoustics coefficients. More difference between minimal and maximal levels - better apparatus. But, what about better features in different parts of frequency range. Resume I don't know technical features to classify apparatus. Because higher distortion level can give better subjective sound perception. Audio apparatus may be finally estimated with proper listening test only.