I set Replay Gain to work on 25 audio tracks. I've taken 5-10 seconds from each track, and compiled them into two downloadable files:
The titles of the tracks, and calculated Radio Replay Gains and Radio Replay Gain adjustments, are shown in the following table.
"Time" refers to the time at which an extract from that track appears in the downloadable audio example.
Time: | Artist/CD - Track: | Replay Gain: | Adjustment: |
0.00 | The Tampera - Feel It (CD single) | 67.0849 dB | -15.9151 dB |
0.04 | Alanis Morissette - Head Over Feet | 71.0121 dB | -11.9879 dB |
0.10 | Celine Dion - Falling into you | 70.5880 dB | -12.4120 dB |
0.19 | Oasis - Wonderwall | 68.0897 dB | -14.9103 dB |
0.25 | Telarc CD80251 - Goldfinger | 75.3096 dB | -07.6904 dB |
0.37 | Telarc CD80183 - Moon River | 87.9934 dB | +04.9934 dB |
0.53 | Telarc CD80221 - Brief Encounter | 77.4630 dB | -05.5370 dB |
1.15 | Elvis - Fever | 80.7907 dB | -02.2593 dB |
1.23 | Oleta Adams - Get Here | 76.0916 dB | -06.9084 dB |
1.32 | HNF+RR test CD 2 - Track 01 (quiet speech) | 92.8158 dB | +09.8158 dB |
1.38 | HFN+RR test CD 2 - Track 03 (pop music) | 80.7753 dB | -02.2247 dB |
1.46 | HFN+RR test CD 2 - Track 19 (classical) | 77.6900 dB | -05.3100 dB |
1.54 | HFN+RR test CD 2 - Track 20 (classical) | 80.7789 dB | -02.2211 dB |
2.01 | HFN+RR test CD 2 - Track 22 (cannon) | 101.1475 dB | +18.1475 dB |
2.05 | Faure - Pavane | 77.8484 dB | -05.1516 dB |
2.17 | 1812 Overture - Mercury Recording | 70.5373 dB | -12.4627 dB |
2.25 | Radio show (music - Basement Jaxx) | 70.7456 dB | -12.2544 dB |
2.33 | Radio show (DJ speech over music) | 69.4228 dB | -13.5772 dB |
2.41 | SQAM CD - Track 50 (Male Speech) | 75.0974 dB | -07.9026 dB |
2.45 | SQAM CD - Track 55 (Haydn Trumpet extract) | 77.2643 dB | -05.7357 dB |
2.53 | SQAM CD - Track 56 (Handel Organ extract) | 75.0690 dB | -07.9310 dB |
3.04 | SQAM CD - Track 57 (Bach Organ Extract) | 91.5487 dB | +08.5487 dB |
3.11 | SQAM CD - Track 58 (Sarasate Guitar extract) | 81.7892 dB | -01.2108 dB |
3.19 | SQAM CD - Track 59 (Ravel Violin extract) | 80.8753 dB | -02.1247 dB |
3.24 | SQAM CD - Track 60 (Schubert Piano extract) | 80.1139 dB | -02.8861 dB |
These tracks were processed by Replay Gain on an individual track-by-track basis, giving the "Radio" level equalisation. The intention is to make all the tracks sound equally loud.
The original SQAM tracks are only extracts themselves, and so do not represent the dynamic range of the entire performance.
I think the results are good (download and compare the Replay Gain version with the original!). However, I hear three problems. Oasis and the 1812 are slightly too quiet for my liking. Also the Bach Organ Extract (small tinkly organ, not the huge organ that precedes it) is too loud (it should be a very quiet instrument). Overall though, the track-by-track "Radio" processing does what you would expect - makes all tracks a similar loudness.
Having heard these results, I decided to specify two Replay Gains. One track-by-track, which causes all tracks to be a similar "Radio" level (appropriate under some listening conditions), and another on a whole CD basis, which is correctable by the user (if the whole CD should be extra quiet or loud). This latter Replay Gain preserves inter-track loudness differences, and would be more suitable for critical "Audiophile" listening.
Finally, the hard limiting on the cannon shot (which was the only track on test which required hard limiting after Replay Gain processing and +6dB Pre-amp gain) changes the sound totally. I don't like it, though many people do! I think if you really wanted to listen to tracks with such a large dyanmic range, you'd just have to turn the hard-limiting and pre-amp off, as suggested for audiophile listeners.
If you find a track which gives a misleading Radio Replay Gain (i.e. a track which doesn't sound the same loudness as all the others after Replay Gain Processing), please contact me.