Clean Vocals vs Consistent Backing Track
Both DeMIX Pro and DeMIX Essentials offer a range of options that allow users to optimize the vocal separation algorithm, depending on the nature of the song to be separated, as well as the use-case for the separation.
Both DeMIX Pro and DeMIX Essentials offer a range of options that allow users to optimize the vocal separation algorithm, depending on the nature of the song to be separated, as well as the use-case for the separation. In particular, obtaining the cleanest sounding lead vocal requires different settings than obtaining a consistent sounding backing track. In this post, I will discuss the automatic vocal separation settings I would use for each of these cases.
Having loaded a song into DeMIX Pro or Essentials, press the blue vocal separation button at the top. This will bring up a menu offering the choice between automatic and guided separation. In this post, I am going to focus on automatic separations only. A later post will focus on guided vocal separations.
The first setting is Source Position. In most cases, the lead vocal is positioned in the center of the stereo field, but in older recordings it can sometimes be positioned to the right or left. This setting should always be chosen to match the position of the lead vocal. When dealing with mono recordings center should be chosen for Source Position.
The vocal width parameter should be chosen based on the characteristics of how the lead vocal was mixed/recorded. The first option, narrow, reflects the case where the vocal is coming from a single point in the stereo field. In some cases, the individual lead vocal tracks are panned to different positions in the stereo field,or alternatively an effect may have been used to spread the vocal. In this case, the vocal can be considered Wide, and the wide setting should be used.
In the case of the vocal reverb extraction parameter, for the best sounding vocal, set this to With Vocal. The setting None should not be used when attempting to get clean, natural-sounding vocals. If you are using DeMIX Pro and wish to have control over the reverb levels on the lead vocal, the setting Separate Track can be used to return the vocal reverb on an individual track.
Vocal Detection Sensitivity typically gives good results when set to low, but it is best to adjust this setting on a song by song basis for best results.
The Snare Reduction parameter is only available in DeMIX Pro. In some songs, too much snare can be left behind in the separated vocal. In this case, turning on Snare Reduction can help reduce the presence of the snare, though at the potential cost of losing some of the plosives and sibilance from the vocal.
Advanced Filtering should be switched on in the vast majority of cases when attempting to get clean vocals.
The parameters Source Position and Vocal Width should be set in the same way as described above when attempting to get a high-quality consistent backing track. Similarly, Vocal Reverb Extraction should be set to With Vocal (or Separate Track in DeMIX Pro if so desired)
Vocal Detection Sensitivity should be set to No vocal detection when attempting to generate a consistent sounding backing track. I recommend this for two reasons. Firstly, on occasion the vocal detector can occasionally make a mistake and leave some lead vocals in the backing track. While this can be overcome using spectral editing in DeMIX Pro, it is impossible to do so in DeMIX Essentials. Secondly, when there is no lead vocal singing occuring in the track, our algorithm typically tracks a single note in a chord being played at that time. The absence of a single note from the backing track in many cases is hardly noticeable in a busy mix.
In DeMIX Pro Snare Reduction should be turned on when a large amount of snare is removed along with the vocals. This will vary from song to song, so it is recommended to separate a short snippet of a song first to hear if too much snare is being removed.
Advanced Filtering should normally be set to on. However, if an extra clean backing track, with little or no trace of vocals is required, advanced filtering can be turned off. It is important to note that this may result in some of the backing track being removed with the vocals, resulting in a lower quality backing track.
I hope you have fun playing with the various parameters in our separation software, and that this post helps you get the results you desire.
Paddy Hallihan2019-03-07 00:00:00