Acquire your test files:
- Select 3 samples from among collection of 24-96 titles acquired from HDtracks. (For this study I chose: Stevie Wonder, "Higher Ground (Innervisions); Jan Garbarek Group, "Tao" (Dresden: In Concert); Allman Brothers Band, "Statesboro Blues" (Live at the Fillmore East). Titles were chosen to have a pool of varied styles and genres.
- Create 16-44 masters from the 24-96 masters using dbpoweramp music converter (or equivalent). This may make the comparisons closer than they might otherwise be due to And some more text.
- similarities between the 16-44 and 24-96 masters that may not occur if they were acquired from independent sources. But I currently own no titles in both 24-96 and CD format.
- Import all file into iTunes (current version: 10.6.0.40).
- Use iTunes to create AAC files from the 16-44 and 24-96 masters.
Now using audio editing software, repeat the following process for each pair of WAV master and AAC copy:
- Import the WAV master under test as one track.
- Invert it (turning all positive samples negative and all negative samples positive).
- Import the AAC copy under test as another track.
- Align the two tracks at a zero crossing adjacent to a recognizable audio peak. (Remember, you are going to have to repeat this step four times per title). Alignment is critical to getting accurate results, and many lossy encoders add silence at the start of a track.
- Mix and render the two tracks into one.
- Post-process the resulting waveform in any way that you feel adds credence to the result. For example, one of my Harman colleagues suggested that I equalize the result with the standard A-weight curve (ANSI S1.4-1981) to emphasize frequencies in the human-audible range. This changed the absolute RMS values in step 7, but did not alter relative correspondence.
- Measure and record the RMS amplitude/volume for the resulting waveform. A lower value indicates closer correspondence. Numbers reported are as measured, not equalized as described in step 6.
No comments:
Post a Comment