What counts as a stable-pitch segment?

A run of consecutive frames where the dominant frequency stays within sixty cents of its neighbours and the amplitude stays above the voiced threshold. Anything shorter than fifty milliseconds is dropped, so brief glitches do not become segments of their own.

Why does a long note sometimes show as several segments?

Because a change of more than sixty cents starts a new run. Strong vibrato, a slide, or a moment where a harmonic briefly overtakes the fundamental can all split a musical note into more than one row, which is worth knowing before reading the count as a number of notes.

What is the amplitude column for?

It carries the mean level of the run on the sample scale, which helps judge whether a tracked frequency is the fundamental or a harmonic. A high frequency at low amplitude usually means the analyser locked onto an overtone rather than the note itself.

Why were no segments found on my recording?

The material is either too quiet to pass the voiced threshold or too noisy for a consistent dominant frequency to emerge. Silence produces nothing by design, and heavily percussive or broadband content rarely holds a pitch long enough to form a run.

Can I export the segment list?

No, the table is a screen-only readout showing the first twelve runs. If the values need to be kept, read them off the table, or cut the interesting section out first so that its runs appear at the top of the list.

Frequency Analyzer

You Might Also Need

Frequency Counter

Detect fundamental frequency

Spectrum Analyzer

Analyze frequency spectrum

Hearing Aid Frequency Reference

Frequency guide for hearing aids

Waveform Visualizer

Visualize audio waveforms in real-time

Frequency Analyzer follows the dominant frequency through a recording and groups it into stable-pitch segments, so a melody, a spoken line or a drifting tone becomes a list of measurable runs. It reports the mean frequency and amplitude of each one.

What is Frequency Analyzer?

The canvas draws each segment as a horizontal bar at its mean frequency, and a table underneath lists the same runs with their start time, end time, mean frequency and mean amplitude.

What each run represents:

The analysed channel is the first one, and the timeline is computed in overlapping 4096-sample frames with a hop of 2048 samples.
Frames are only considered voiced above an amplitude threshold of 400 on the sample scale, which keeps silence from being reported as a very low note.
Segment merging tolerates a sixty cent deviation between neighbouring frames, so a run breaks when the pitch genuinely moves rather than on every small fluctuation.
Segments shorter than fifty milliseconds are discarded, and the table shows the first twelve runs.

Between the frame level and the table there are two stages. The first produces a dominant frequency per frame together with its amplitude, and the second walks that sequence and starts a new segment whenever the frequency jumps beyond the tolerance or the amplitude falls below the voiced threshold. The sixty cent allowance is a musical choice: it is slightly wider than the range a steady note wanders through as vibrato and intonation move, yet far narrower than a semitone, so a real change of note always starts a new segment. The amplitude column is useful in its own right, because a run that is high in frequency but low in amplitude is usually a harmonic slipping past the fundamental, while a strong low run suggests the fundamental has been tracked properly. Presenting the findings as a segment table rather than a single number is what makes the tool usable on real material, because it keeps the stable-pitch segments separate and shows how long each one held. A melody read this way becomes a series of measurable runs, and an unsteady delivery becomes a long list of short entries rather than one average that describes nothing in particular, which is exactly the kind of problem a single averaged pitch would hide.

What the analysis cannot promise:

Only the first channel is analysed, so a stereo file is summarised by one side.
Silences and unpitched material produce no segments rather than a measured noise floor.
The canvas stops showing what is beyond the loudest voiced segment scaled up by ten per cent, so unusually high runs are clipped at the top of the plot.
Only the first twelve segments are listed, however many were found.
The reported frequency is a mean over each run, so a slow glide is summarised as one number instead of a shape.
There is no export of the segment list.

Use it for material that is meant to be pitched: a tuning check on a sustained note, a spoken take where you want to see how wide the delivery's range actually is, a recording of an instrument to check whether the low notes are coming through, or a drift test on a synthesiser. For a single number instead of a timeline, the frequency counter reports one fundamental with a confidence figure, and the spectrum view is better for judging an instrument's harmonics.

How to use Frequency Analyzer

Drop an audio file onto the panel and wait for the segment analysis to finish.
Scan the canvas for runs, reading height as mean frequency and width as how long the pitch held.
Read the table for exact start and end times, mean frequency and mean amplitude of each run.
Compare the reported range against what you expected from the material to spot missing fundamentals or drift.

When to use Frequency Analyzer vs related tools

Frequency Analyzer is built for pitched material where the dominant frequency moves over time, which makes it the natural check on tuning, range and drift. Frequency Counter condenses the same kind of measurement into one fundamental and a confidence percentage, and Spectrum Analyzer shows harmonic balance without a time axis. For unpitched problems such as hum or noise, Spectrogram Viewer is the clearer view.

Privacy & Security

This tool runs entirely in your browser — no data ever leaves your device. There is no server round-trip, no upload, no logging, and no account required. Your input is processed locally using client-side JavaScript and is never stored, transmitted, or accessible to anyone else. When you close the tab, everything disappears.