HERTZKIT / PRACTICAL GUIDE
Understand Practice results and device latency
Understand cents, unread notes and latency in HertzKit Practice. Learn to repeat a phrase with comparable measurements and a clear practice goal.
On this page6 sections
By Marcelo Borasi · Updated November 2026
HertzKit Practice compares a one-note-at-a-time phrase with a target you confirm. It does not grade talent; it offers approximate readings to guide practice.
A meaningful target
Repeat the same passage twice with the same microphone and distance. Listen to both takes before checking the measurements. If the difficult notes move between takes, try again with less noise or a shorter phrase.
Start with the example or a short solo phrase. Listen to the reference notes. Review audio-derived notes: chords, noise, vibrato and unclear attacks can confuse detection. An incorrect target creates an incorrect comparison.
Read pitch by note
The chart summarizes each measured note. Zero is the target; above means higher and below means lower. A semitone equals 100 cents. Typical deviation is the median of the absolute note-center offsets from their targets, not a musical-quality score or a demand to remove vibrato.
“No stable reading” means the system lacked sufficient evidence, not that you performed badly. Check microphone, noise, register and speed before changing your performance.
Timing and latency
Capture and playback delay can shift estimated onsets. Rhythm comparison is optional and also depends on analysis windows: displaying milliseconds does not guarantee that precision. Compensation adjusts how the take is compared; it does not speed up the device or correct the original recording.
- Choose and confirm a melody in Practice.
- For onset comparisons, open the optional acoustic measurement. Use moderate speakers and a nearby microphone; do not sing or play.
- The tool plays three tones and seeks a consistent return. If unsuccessful, it retains the previous latency value.
- The number includes speaker, air and microphone delay, not separate measurements of each component. Repeat after device changes.
- Return to headphones for practice. Speaker calibration is indicative and may differ from this new configuration.
Practice one difficulty
Isolate the note with the largest reliable deviation, listen and repeat more slowly. Then return to the full phrase. Increase tempo when comfortable, not because a number demands it.
Keep your progress
Optional local storage does not contain audio. Named copies are independent of automatic saving. Export JSON to move exercises; download a recorded WAV separately. Other users of a shared device may access local data.
The PNG/PDF report retains per-note readings, including missing measurements. Including a comparison requires complete takes from the current session and confirmation of comparable conditions. It does not use imported histories to claim progress. Review the preview before sending it to your teacher. How to prepare and share the report.
References and scope
This guide describes Practice controls and limits; it is not pitch certification or artistic assessment. Web Audio documentation explains browser output latency, but that figure does not cover the entire microphone and speaker path. MDN: AudioContext.outputLatency
From a listening goal to a portable practice
Before recording in Practice, choose one goal: matching pitch, arriving at a note or listening for continuity. Record the full phrase, work on one part and repeat the original phrase at the same speed. A listening goal does not produce a grade.
Compare by ear first. A pitch difference is shown only when the readings and capture settings permit it and you confirm the same conditions. Noise, register differences, clipping and unstable readings can prevent a useful comparison. It does not assess expression, diagnose technique or establish why a result changed.
Download .hksession to preserve the recreated reference, both takes and your observation. Open it in Sessions for comments and feedback. The takes are imported audio there, without a numerical comparison. A TXT summary does not include audio. Download before starting the optional variation, which begins a separate comparison.
Open guided practice →