SergeSE
Reviewer at SoundExpert.org
- Joined
- Apr 10, 2014
- Posts
- 47
- Likes
- 28
Some of SoundExpert internal research could be interesting for Head-Fi people, so I would like to share the selected results. We investigated performance of audio section (DAC+analog circuits) of several portable devices. In particular we compared initial waveform in input file (44/16) with output waveform of signal recorded (96/24) from phones output of devices loaded with 33 Ohm resistors. Playback levels of devices were matched. We used 11 signals for comparison: sine wave, white noise and nine real-life sound samples. The signals were divided into 100 ms time blocks and corresponding input/output blocks were compared with each other. Result of comparison is measured with difference level parameter (Df) in dB: if waveforms of the blocks are identical Df = -Inf, if waveforms are completely different Df = 0. In order to visualize the differences between signals Df values were color coded according to this color map:
http://soundexpert.org/documents/10179/130750/df_colormap.png
Each 100 ms block of signal is represented by 1x100px bar; sequence of such bars (diffrogram) represents the whole signal. For the ease of signal recognition the diffrogram is combined with spectrogram, where amplitudes of spectral components are coded with lightness/darkness of color.
Below are diffrograms of four devices:
FiiO E17 - USB DAC headphone amplifier
http://soundexpert.org/documents/10179/130750/diff_FiiO_E17.png
iPhone 5s - smartphone
http://soundexpert.org/documents/10179/130750/diff_iPhone_5S.png
iPhone 6 - smartphone
http://soundexpert.org/documents/10179/130750/diff_iPhone_6.png
iriver E100 - multimedia player
http://soundexpert.org/documents/10179/130750/diff_iriver_E100.png
Figures at the bottom are difference levels (Df) computed for corresponding signal parts. As you can see all devices show almost equal performance with sine signal but white noise and real-life signals are transferred through them with different accuracy. Also iPhone 5S and iPhone 6 have exactly the same audio section, their diffrograms are identical (small deviation of Df values are due to imperfection of measurement procedure). Thus, diffrogram is a simple and visual way of imaging signal degradation in various audio circuits.
In case of interest I'm ready to prepare and publish here diffrograms of some other devices for which we have recordings. Details of diffrogram computation are in the article “Diffrogram: visualization of signal differences in audio research”.
http://soundexpert.org/documents/10179/130750/df_colormap.png
Each 100 ms block of signal is represented by 1x100px bar; sequence of such bars (diffrogram) represents the whole signal. For the ease of signal recognition the diffrogram is combined with spectrogram, where amplitudes of spectral components are coded with lightness/darkness of color.
Below are diffrograms of four devices:
FiiO E17 - USB DAC headphone amplifier
http://soundexpert.org/documents/10179/130750/diff_FiiO_E17.png
iPhone 5s - smartphone
http://soundexpert.org/documents/10179/130750/diff_iPhone_5S.png
iPhone 6 - smartphone
http://soundexpert.org/documents/10179/130750/diff_iPhone_6.png
iriver E100 - multimedia player
http://soundexpert.org/documents/10179/130750/diff_iriver_E100.png
Figures at the bottom are difference levels (Df) computed for corresponding signal parts. As you can see all devices show almost equal performance with sine signal but white noise and real-life signals are transferred through them with different accuracy. Also iPhone 5S and iPhone 6 have exactly the same audio section, their diffrograms are identical (small deviation of Df values are due to imperfection of measurement procedure). Thus, diffrogram is a simple and visual way of imaging signal degradation in various audio circuits.
In case of interest I'm ready to prepare and publish here diffrograms of some other devices for which we have recordings. Details of diffrogram computation are in the article “Diffrogram: visualization of signal differences in audio research”.
