There is another program that I find very useful when producing learning tracks for Barbershop songs. That program is Harmony Assistant by Myriad Software. Harmony Assistant, when used with the Virtual Singer add-on, is capable of producing very nearly human-sounding vocal tracks. When the notes are adjusted to the proper Just Intonation tuning using JustChords, the result is a great starting point for learning a new Barbershop song.

I have written a Harmony Assistant plug-in, available here, that imports a MusicXML file which has been augmented with tuning info by JustChords and produces a 4-part song with lyrics, ready for splitting into separate learning tracks for a quartet or chorus to use. Here is a brief outline of the process:

• Create a MusicXML file of the song. One free program that is quite capable of producing a usable file is MuseScore. This file can be a "standard" two-staff file with Tenor and Lead on one staff and Baritone and Bass on another staff, or it can be a 4-staff file. It should include the lyrics.
• Open the MusicXML file with JustChords and step through it, making chord selections and tuning tweaks to perfect the tuning.
• If the file is of the two-staff variety, select Split Staves from the File menu to create a four-stave version.
• Save the file. It will now contain tuning information for each note.
• In Harmony Assistant, select File->Import->Import MusicXML With Pitch Bends (near the bottom of the menu) to open the augmented file.
• Enable Virtual Singer, choose "Virtual Robert" for each voice, select US English as the language, tweak the voice parameters to minimize vibrato and scooping (see below), and play the song. You probably will need to add some pronunciation tweaks to get the words to sound more natural.
• Create separate tracks for each voice. You can do this by soloing one voice and exporting it as an MP3.
• Combine the separate vocal tracks using Audacity or any other Digital Audio Workstation tool. This step can be done using just Harmony Assistant, but I find Audacity easier.

The process is actually pretty simple, and most of the necessary tools are free. Harmony Assistant plus Virtual Singer currently (May 2020) cost only $110 for a lifetime license, including all future upgrades. The other tools mentioned here are absolutely free. With a little practice, you should be able to go from a PDF to perfectly tuned learning tracks for a new song with less than 8 hours of work (estimated).

VirtualRobert

Virtual Singer comes with many different sampled voices. I find that the one called "VirtualRobert" works quite well for Barbershop, once you tweak it a little. The changes I recommend are the following:

• Open the Virtual Singer window. Drag the four little "paper dolls" to four separate spots on the "stage" so you can select each one. This step also sets the "pan" position (left to right) and relative balance (nearness to the front of the stage).
• Click on the flag icon below each singer and select "US English." The flag icon will show the American flag.
• Double-click one of the singers. This opens the "Edit" window. Select "VirtualRobert" from the list.
• Click on "Edit voice." In the window that pops up should be five tabs. Click on the "Vibrato" tab.
• Set the two parameters, "Rise time" and "Depth" to 0 by dragging the knobs with your mouse.
• Click on the "Pitch" tab. In the lower right, under "Drift F0:", set the "Min time" and "Max time" values to 0 by dragging the knobs with your mouse.
• Click on "Ok". Click on "Add to list" to save your modified voice to the list of available voices. Choose a new name; I chose "VirtualRobert-BBS".
• Use the "<<" or ">>" buttons to select another singer. Click on your newly-added voice in the list. Repeat for the other voices.

If you find zero vibrato and zero scoop to sound too "robotic," feel free to add some humanness back in.

SAMPA

You can get Virtual Singer to pronounce a syllable completely differently from the way it's spelled. I find that I need to "correct" a handful of syllables in nearly every song. The mechanism for doing this is called SAMPA. This notation gives you precise control over the consonants/vowels that Virtual Singer uses to render a given syllable, if its pronunciation is somewhat off. The SAMPA notation does not show up in the printed score, so you can still use the same file for printing, if desired, after correcting the pronunciation.

Help Topic Index