Tuning, voice and ear training
HertzKit Practice
Listen to a short phrase and repeat it with your voice or instrument. Then listen to your take and try again.
First visit: try the example. The microphone opens only when you record or test the input.
Microphone and level
Use the automatic input or choose your microphone. The level setting affects future recordings.
Microphone level in HertzKit
100% preserves incoming amplitude; 50% halves it (−6 dB). Adjust from 0 to 200% (+6 dB), including while the microphone is active. Meters show approximate peaks from −60 to 0 dBFS, not acoustic volume. This control changes the signal inside HertzKit, not the hardware gain. If the incoming signal is already distorted, lower the system or interface input. Increasing gain also amplifies noise. This setting is not saved when the tab closes.
Audio input
Device names appear after microphone permission is granted.
1. Choose your melody
Choose the example if this is your first time. For audio, MIDI or microphone, open “Use my melody”.
You’ll hear C, D, E, G, E, C. Sing or play one note at a time, without accompaniment.
Use my melody · audio, MIDI or microphone
Choose a recording of a solo voice or instrument. Then select a phrase of up to 20 seconds to create your practice reference.
Confirm permission above to choose your audio file.
Choose the phrase to practice
Drag Start and End on the waveform with a mouse or finger. Selection: 0.2 to 20 seconds. Waveform scale only changes how much of the file you see.
Excerpt tempo
Estimation uses 5–20 seconds. You can also enter the BPM.
MIDI contains notes, not audio. Up to 1 MB; one melodic track without chords, sustain or pitch bend.
First record a phrase you want to repeat. This creates the reference; it does not evaluate your performance yet. Up to 20 seconds, without accompaniment.
Press “Finish phrase” when done. Review the detected notes in the next step.
2. Review and listen to your melody
Listen to the recreated reference. If you used a recording, compare it with the original before continuing.
Recreated reference · notes played with the selected instrument. This is not your original recording.
Select a note to hear it. Change octave if needed.
A lower number plays more slowly. Start at a comfortable speed.
Original audio · your recording
The practice reference uses estimated notes, not the original recording. Check differences before continuing.
You can correct notes and timing in the optional editor.
Edit notes or choose a phrase · only if needed
Start and duration are in beats: 1 beat is a quarter note. Middle C = C4. No overlapping notes. Scroll the table on mobile.
No notes.
| # | Note | Start | Duration | Remove |
|---|
Tuning and rhythm variations · advanced
Leave this at 440 Hz unless you need another tuning. Variations use editable rules, not generative AI.
Your practice
Microphone closed
Your latest take
- Put on headphones and find a quiet place.
- Use the record button and allow the microphone.
- Wait for 3, 2, 1. Start when “Recording” appears. Recording ends automatically.
Use one voice or instrument. With speakers, the microphone may capture the guide instead of your performance. Artistic quality is not evaluated.
Latest recording and measurements
This report contains one take. At the end of the journey, TXT keeps the comparison text; .hksession keeps the recreated reference and both recordings for Sessions.
Measurements for this take
Listen to the recording first. Then check what could be measured and choose what to repeat.
View readings and chart by note
Pitch by note
Zero is the target note. Positive values are higher; negative values are lower. Gaps indicate uncertain measurements. This chart shows note summaries, not a continuous pitch trace.
| Note | Pitch | Onset | Analyzable signal |
|---|
100 cents equals one semitone. “No stable reading” means the note could not be measured; it is not your mistake. The percentage describes analyzable signal, not correct notes.
A stable reading can confuse an octave or harmonic. Listen to your take before drawing conclusions; these numbers do not certify performance quality.
Save or share a visual report · optional
Prepare a report for your teacher or to keep a practice record. The preview contains readings, not audio. Sharing is optional and files are generated on your device.
File preview
On small screens you can scroll the preview. Open the downloaded PNG or PDF to zoom in. PDF uses a light background for printing.
Review the title and observation before sending. Your name, original file name and audio are not included automatically. “Share image” appears only when your browser supports it. It opens the device sharing panel. You can also download the PNG and send it with your app. No public link to your results is created.
Timing, progressive tempo and MIDI export · advanced
Onsets are approximate and can include device delay. Latency is not measured automatically.
Estimate device latency · optional
Temporarily use speakers at moderate volume with the microphone nearby. Stay silent: HertzKit plays three short tones and estimates their acoustic return delay. Headphones cannot be measured this way. This includes speaker, air and microphone delay and is only an approximate compensation for this setup.
A consistent measurement updates the latency field. Recheck after changing devices. Return to headphones before practicing; a failed measurement does not change the field.
MIDI keeps notes and timing. Its sound depends on the player.
Help and practice options
What to save and where to continue
- WAV takes: download them to keep the audio. Closing the tab can lose them.
- Completed practice · .hksession: includes a recreated reference, the first and final recordings and your observation. Open it in Sessions without an account. Download before starting a different phrase or closing the tab.
- My exercises: saves notes and settings, not recordings.
- TXT summary or PNG/PDF report: keeps the information described by each download; no audio is included.
To plan your next session, open Practice Projects in another tab. Record your goal or observation there: takes and measurements are not transferred automatically.
How to use Practice and understand the readings
Choose a short melody you can sing or play, confirm the reference notes and test input before capturing a take. Repeat the same phrase with the same options to compare pitch and timing. If notes have no reading, listen to the recording: missing measurements do not prove you did not play.
Supported material and measurement limits
One-note-at-a-time phrases from C2 to C6: up to 64 notes, 32 beats and 8 beats per note. Audio file up to 10 minutes and 128 MB; only the selected phrase, up to 20 seconds, is analyzed; MIDI up to 1 MB. Audio tempo can be suggested when a clear pulse is found; check it by ear. Meter is not detected automatically. It does not separate vocals or evaluate complete songs.
Noise, chords, wide vibrato and fast notes can prevent measurement. Notes shorter than 0.2 seconds at the chosen tempo should be practiced more slowly. The detector proposes notes for review; it does not certify transcription or evaluate musical intention.
To record, allow microphone access in a compatible browser. It is released when you finish or cancel. Switching apps or locking the phone cancels an incomplete take. Timing readings are approximate and may include device delay.
Analysis and practice variations run on your device using the selected notes and passage. Files and takes are not uploaded to HertzKit.
Files and progress
Up to 10 minutes and 128 MB per file, mono or stereo. Both limits apply. Supported formats and available memory depend on your browser and device; if you have trouble, try a shorter excerpt.
Loading and analysis may have several stages. If no percentage is shown, wait for completion or cancel. After downloading, open the file to check that it was saved correctly.
Frequently asked questions
Does the percentage decide whether my performance is good?
No. It describes comparison with that reference and the data that could be measured. Noise, vibrato, attacks and latency may affect results. Expression, intent and artistic quality need listening and context.
Does changing input level modify earlier takes?
No. It affects subsequent captures and the signal analyzed at that moment, not already loaded files. Download the latest WAV to keep audio; saved exercises contain notes and settings, not recordings.
Keep an activity together
After completing your practice, download the recreated reference and both takes as .hksession. Open that copy in Sessions to listen and add feedback. Saved Practice exercises still keep only notes and settings.