Voice Trainer: practise every voice on its own — straight from the score

Anyone who has tried to learn a new voice on their own at home knows the problem: the piano demo plays, all voices sound at the same time, and your own line gets lost in the texture of the others. This is exactly the moment we built the Voice Trainer in Chorilo for. From a single MusicXML file of a piece, the system automatically generates a separate exercise for every voice — soprano, alto, tenor, bass or any other voice part in your choir. Your singers open the exercise in the mobile app, pick their voice, adjust the volume of the other voices through a mixer, slow the tempo down if needed and can even turn on the microphone so that Chorilo can tell whether they are hitting the right notes. A long stretch between rehearsals turns into independent practice with feedback — and a tricky passage into a few targeted repetitions instead of frustration.

What the Voice Trainer in Chorilo actually does

At its core the Voice Trainer is an automatic vocal coach: you upload a MusicXML file on the detail page of a piece of sheet music, and the system renders multiple audio stems per voice, an animated music notation and a timeline of every single note from it. After that, your members find their own voice exercises per piece in the Chorilo app — one for soprano, one for alto and so on. Each exercise contains a separate audio track per voice in three tempi: original tempo, 75 percent and 50 percent. So a beginner soprano can sing along at half the final tempo first, while an experienced bass tackles the same rehearsal at full speed straight away. In addition Chorilo generates a metronome track that runs in the mixer and quietly sets the beat.

That makes the Voice Trainer more than a digital music stand and more than a simple audio player. It is a fully fledged practice tool for choral music that closes the gap between handing out the score and the rehearsal — exactly the phase where most time is lost today. Anyone who has ever spent a rehearsal going through the same passage in the soprano line for the fourth time knows what an hour of focused preparation at home is worth.

How practising in the mobile app works

On a smartphone or tablet, a member opens the score in the sheet music area, taps the microphone icon at the top of the toolbar and lands in the Voice Trainer. The exercise loads immediately and the member sees all voices of the piece as a multi-voice score with a cursor that runs along during playback. Through the voice dropdown they can switch between soprano, alto, tenor, bass or any other voice present in the file at any time, without reloading the exercise. Chorilo remembers the chosen voice locally per member and per exercise, so on the next opening the most recently practised voice is preselected automatically.

The tempo can be switched through another menu item — preset to 100 percent, with 75 percent and 50 percent next to it. These tempi are not just playback speed but real, freshly rendered recordings without pitch distortion, so even half-tempo does not sound artificial. Anyone who wants to practise just one passage over and over marks loop points directly in the score view and lets the Voice Trainer repeat the section automatically — very handy for a tricky modulation or a multi-bar entry.

The real trick is the voice mixer. Behind its own icon, every voice present in the file gets a volume slider and a mute switch. Anyone who only wants to hear their own voice mutes all the others and gets a clean solo track. Anyone who wants to listen into their voice while still hearing the others quietly in the background turns them down to twenty or thirty percent. The mixer also contains a track for the metronome that can be switched on or off independently of the voices. All these settings are also stored locally per member and per exercise, so each singer finds their preferred mix straight away the next time they practise.

The Voice Trainer becomes really special when the member turns on the microphone. Behind a small mic button sits pitch detection that analyses the sung tones in real time and compares them to the score. While singing, a coloured marker moves to the actual sung pitch — green when the note is spot on, yellow for a slight deviation, red for a semitone or more off. At the end of the exercise a friendly evaluation appears as a coloured graphic in red, yellow and green that shows at a glance how well the run-through went. This creates motivating feedback without pressure and without anyone else judging — feedback that gives a good sense of what already works and where there is still room for improvement next time. Anyone who has shifted the octave — a bass voice trying out a tenor line, or vice versa — can switch on an option that automatically counts octave displacements as correct. The microphone is optional: anyone who simply wants to read along and listen quietly does not turn it on, and Chorilo neither accesses the microphone nor records anything.

How an exercise is created: upload via the web

A Voice Trainer exercise is not created in the app but in the web frontend — it is a choir director's job, because the MusicXML file of the piece is needed for it. Open the detail page of a piece of sheet music at www.chorilo.com, scroll to the Voice Trainer exercises section on the right and click on "Upload exercise". You enter a title — usually the piece name or a variant of it — and an optional short description, such as "concert version" or "with repeat of the middle section". Then you select the MusicXML file from your computer. Chorilo accepts the extensions .musicxml, .xml and .mxl up to a size of ten megabytes; this covers practically every piece you have created in MuseScore, Sibelius, Dorico, Finale or another common notation program.

As soon as you have clicked on "Upload", the system takes over. The exercise status first jumps to "Pending" and shortly after switches to "Processing". In this phase Chorilo creates the necessary practice recordings for every voice in the three tempi from your file and prepares the score view for the app. Depending on the length of the piece this usually takes one to five minutes. You can safely close the browser window — processing continues in the background, and the next time you open the sheet music detail page the status shows whether the exercise is ready. As soon as "Done" appears, the exercise is immediately available in the app for every authorised member.

What the MusicXML file has to contain

The quality of the resulting exercise stands or falls with the quality of the MusicXML file. In daily life this is rarely a problem — modern notation programs export clean MusicXML — but a few points are worth checking before uploading. Each voice in the piece should have its own track (in MusicXML jargon a "part"), with its own voice name and its own lyrics. Chorilo creates a separate exercise from every part; a shared track for soprano and alto with two parallel lines therefore works less well, because the system would treat them as one voice. Clear, unambiguous voice names like "Soprano", "Alto", "Tenor" and "Bass" help your members pick the right exercise; a plain "Voice 1" is accepted but is less meaningful for the singers.

A tempo marking in the piece is also important, because without it Chorilo cannot determine the original tempo. Ideally place a marking at the beginning of the piece such as "quarter = 100" or a comparable indication. From this original tempo, versions at seventy-five and fifty percent are then calculated automatically. It is also important that repeats are written out in the file. Repeat barlines, voltas and Da Capo instructions are not unfolded by the Voice Trainer — if a middle section is meant to be sung twice, it has to appear twice in the MusicXML. Most notation programs offer a function for "writing out repeats" or "Unfold Repeats" which you apply once before exporting. Chord notes — that is, several pitches at the same time in the same voice — are treated as a single sound and are not split into separate voices. So if your soprano divides into two lines at one point, these should be set up as two separate parts so that each line gets its own exercise. For pieces with complex notation — for example with ossia voices, many cue notes or extended notation forms — it is worth taking a look at the file before exporting and hiding everything that is not meant to be sung.

Replace, reprocess or delete a file

It happens that an exercise is not quite right after the first upload — perhaps a voice was renamed in the notation program, a tempo marking was added or a repeat was inserted. For exactly these cases there is the "Replace file" button next to the exercise. You simply select the new MusicXML file, and Chorilo discards the old audio files, uploads the new version and starts processing from scratch. Title and description are kept, so you do not have to take care of the metadata again. As soon as the status switches back to "Done", your members automatically see the new version in the app — no push and no member-side action is needed here either.

Should the automatic processing fail once on a complex file, the status jumps to "Failed" and shows a short error message. In most cases a click on "Reprocess" is enough to restart the job — for example when a temporary bottleneck disturbed the first attempt. If that does not help, it is worth briefly checking the MusicXML file, making a small correction and uploading the file again via "Replace file". An exercise you no longer need can be removed via "Delete"; all related audio files are automatically cleaned up at the same time.

What else to keep in mind when creating exercises

Voice Trainer exercises are managed per piece of sheet music — not per event and not per member. So anyone who has the right at ensemble level to edit sheet music can also create, replace or delete Voice Trainer exercises. Members with read access to the sheet music automatically have access to all related exercises. There is deliberately no separate permission for the Voice Trainer, so that a single piece and its exercises never drift apart. Licensed scores from the Chorilo sheet music shop come with Voice Trainer exercises, where available, already prepared; as a choir director you no longer need to upload anything there.

For the piece itself, the rule of thumb is: the cleaner the source, the better the result. A short file with clearly named voices, a tempo marking and an unambiguous lyric assignment per voice is processed in under a minute and yields an exercise in which the score, the audio and the cursor run synchronously. For longer works with many repeats and voltas processing takes a little longer, but the result is just as precise. If you are unsure whether a particular piece runs through cleanly, an initial test with a short trial file often helps more than long theorising — thanks to the lean processing, a trial run costs little time.

If something doesn't work: we help directly

The Voice Trainer works fully automatically, and normally a single upload is enough for your singers to start practising right away. If you do get stuck somewhere, however — be it because processing repeatedly fails, a particular voice is not detected or the result in the app does not sound the way you expected — just drop us a short message at support@chorilo.com. Ideally send the MusicXML file along and a few words about what you wanted to achieve and what does not work. We look at it personally, usually reply within a day and rework many special cases directly in the background, so that your file runs cleanly on the next attempt. We are also happy to receive suggestions for new features around the Voice Trainer at this address — for instance additional tempi, transposition or better feedback for special notation forms.

Why the effort is worth it

Practising voices on their own is one of the most effective methods of getting to a confident choral result quickly. Until now this often failed in practice: rehearsal CDs could hardly be updated, MIDI files sounded unpleasant, and manually recording a practice break cost choir directors hours. With the Voice Trainer in Chorilo, the effort turns into a single upload, practising at home turns into a self-directed learning process with feedback — and the concert turns into a result where every voice arrives at rehearsal prepared. Anyone who has introduced this feature once rarely wants to give it up again. Just try it on your next piece: a few minutes for the upload, a short note to the choir, and the Voice Trainer is up and running.