"Audioscanner" and "Advanced Audio Equalizer" are VST plugins that manually assist realtime musical audio signal separation in stereo recordings by using panning, phase alignment and other magnitude cues in a STFT scheme. Their theoretical background is described in the following article published at the AES 120th Convention (2006):
Demixing Commercial Music Productions via Human-Assisted Time-Frequency Masking.
Jordi Bonada, Alex Loscos, MarC Vinyes Raso.
Matlab demo of the algorithm with fixed panning and phase difference ranges : matlab_demo.zip
The synthesis method is also compared to the classical synthesis method of Daniel W.Griffin and Jae S. Lim, Signal estimation from modified short-time fourier transform. IEEE Transactions on Acoustics, Speech and Signal Processing, 2, 1984.
The Beatles - Help
Louis Armstrong - What a Wonderful World
In the first part, we listen to the original song.
Next two filters of the plugin are activated in order to separate the voice from the background. The first one filters by phase difference. The second one, filters the remaining sound by panning. The vocals seem to have been recorded with a mono microphone without adding artificial stereo reverb and they are panned to the center, so they share the same phase in both channels and the whole process leads to a fairly good separation.
The separated voice has a high quality. It sounds reverbered, but listening carefully to the mixture, we realize that the voice is already reverbered in the mixture. This reverberation was probably caused by the room where Louis Armstrong recorded the vocals using a mono mic. This is one example of the utility of this plugin to analyze recordings and learn details about the recording processes of your favorite songs, as well as focus on the personal interpretation of Louis Armstrong.
Finally, in the last part, we pursue voice removal. We switch to "mute" both filters and, we set to 0 the values of the region where the voice is present.
Pearl Jam - Better Man
We copy the original song into two tracks.
In the first track, we activate two filters of the plugin to obtain the guitar. The first one filters by panning the zone where the guitar is located (left). Next a low-pass frequency filter is applied to remove the vocal residual noise. At this point we can listen only to the guitar. This is already a good result for a guitarrist who is trying to do musical analysis of the guitar, but we will go on and perform a remix of the song.
In the second track, we activate one filter of the plugin in "solo" mode selecting the complementary of the previously selected panning zone. Following this procedure, we are able to extract everything but the guitar.
Now that we have one track with only the guitar and one track with the rest of the song instruments, we apply a distorsion plugin to the guitar track after some seconds of clean guitar. A personal remix of the song is obtained. We selected a distorsion effect but any other effect might be chosen.
MarC Vinyes Raso, Jordi Bonada.
MarC Vinyes Raso, Oscar Mayor, Vasileios Pantazis.
Use our contact page if you wish to contact us.