Voice trainer (AI vocal track)

What is this?

The voice trainer is an AI-powered feature that automatically extracts individual voice tracks from any recording: soprano, alto, tenor and bass. That way you can hear your own voice clearly in the foreground — the other voices stay quieter but remain available as orientation.

Perfect for practising:

• Difficult passages with your own voice clearly in the foreground
• Learning a new voice from scratch
• Self-test: check whether you are confident by muting your voice and singing along

How to use the voice trainer

1. Open the piece via the sheet music overview.
2. Switch to the Voice trainer area on the detail page.
3. If no tracks exist yet: start processing via Calculate voice trainer. Pick the source file (audio or MusicXML).
4. Wait for processing (1–5 minutes). You get a notification as soon as it is finished.
5. Pick your voice and adjust the volumes:
   • Own voice loud for learning the melody
   • Own voice quiet for self-test
   • Other voices medium for orientation

What you find in the voice trainer

• Mixer per voice with its own volume slider
• Solo button per voice — hear only this voice
• Mute button per voice — mute this voice
• Tempo slider — practise slower, test at full speed
• Loop function — repeat difficult passages on a loop
• Jump markers — jump straight to important spots

What the AI does

The AI analyses the uploaded audio or MusicXML file and separates the individual voices from each other. With clean a cappella recordings this works almost perfectly. With complex recordings featuring accompaniment or reverb, small overlaps may remain — that is normal and barely affects practice.

Permission

Listening is open to every member with sheet_music.view. To start processing or change the source file you need sheet_music.edit. That way you save processing capacity for pieces that really need it.

Tips

• Start the voice trainer once per piece — the tracks are kept permanently and do not need to be recalculated each time.
• Pick the cleanest source file possible — a MIDI rendering or an a cappella recording delivers significantly better results than a noisy live recording.
• Explain the mixer to members: solo, mute and volume are the three core tools.
• Use the tempo slider when learning — 80 % tempo is often ideal for the first run-throughs.