Voice Quality Report (Jitter, Shimmer, HNR)

Drag & Drop or Click to Select a Sustained Vowel

A held 'ah' of a few seconds, one voice, close to the microphone

Tip: paste an audio file from clipboard (Ctrl/⌘+V).

Jitter (local)

—

Period-to-period variation. Praat's pathology threshold is 1.04%.

Shimmer (local)

—

Amplitude variation. Praat's threshold is 3.81%.

Harmonics to noise

—

Above 20 dB is a clear sustained vowel; a breathy or hoarse voice sits under 10.

Shareable link

Settings are written to the URL as you change them. Nothing differs from the defaults yet.

Oxyd To put the result on a track with other audio, Oxyd is a multitrack editor with the same effects. Open Oxyd

The acoustic voice report. A sustained vowel is a train of glottal pulses, and almost every clinical and research measure of voice quality is a statistic of how regular that train is: how much the period wobbles, how much the amplitude wobbles, and how much of the sound is not periodic at all. The page marks the pulses and computes the measures Praat’s voice report prints, with its definitions.

Pitch and pulses

Each 10 ms frame gets a pitch by normalised autocorrelation, with Praat’s octave cost so a 150 Hz voice does not read as 75. Inside voiced stretches the pulses are marked one period at a time, each at the waveform’s largest peak within a window around where the last period predicts the next. Consecutive periods are kept only where they stay within a factor of 1.3 of each other and their amplitudes within 1.6, so a break in voicing does not count as a huge jitter.

Jitter

Local jitter is the mean absolute difference between consecutive periods divided by the mean period. RAP compares each period with the average of itself and its two neighbours, PPQ5 with the average over five, which smooths out a slow drift in pitch and leaves the cycle-to-cycle roughness. The absolute figure is local jitter in microseconds.

Shimmer

The same three shapes on the peak-to-peak amplitude of each period: local, and against 3, 5 and 11 point averages, plus local shimmer in decibels, the mean absolute log ratio of consecutive amplitudes.

Harmonics to noise

For each voiced frame the strength of the autocorrelation at the pitch period, r, becomes 10 log10(r / (1 - r)), and the report is the mean over frames. It is a ratio of the periodic energy to everything else, so breathiness, hoarseness and a noisy microphone all lower it.

Settings

The pitch floor and ceiling bound the search, at Praat’s defaults of 75 and 600 Hz. Raise the floor to 100 Hz for a woman’s voice and the ceiling for a sung soprano; a floor set too low lets the search find half the pitch.

Reading the result

Every measure describes the recording as much as the voice. A noisy room raises HNR’s noise and every perturbation with it; a cheap microphone’s hiss does the same; a compressed file adds its own. Compare recordings made the same way, and treat the thresholds as what they are, cut-offs for one recording protocol rather than diagnoses.

Frequently Asked Questions

What do jitter and shimmer measure?

Jitter is how much the length of each vocal fold cycle varies from the one before it, as a fraction of the average cycle; shimmer is the same for the cycle's amplitude. A steady voice has both near zero. Praat's thresholds for pathology are 1.04 percent jitter and 3.81 percent shimmer, on a sustained vowel recorded cleanly.

What should I record?

A held vowel, an ah or an ee, for three to five seconds at a comfortable pitch, close to the microphone in a quiet room. Speech works for the pitch statistics but the perturbation measures are defined on sustained phonation and read high on anything else.

What is HNR?

The harmonics-to-noise ratio: how much of the sound is the periodic part against how much is noise, in decibels, from the normalised autocorrelation at the pitch period, averaged over voiced frames. Above 20 dB is a clear vowel; a breathy or hoarse voice sits under 10.

Do the numbers match Praat's?

The definitions are Praat's, including the 1.3 period factor and 1.6 amplitude factor that decide which cycles count, and the autocorrelation pitch with its octave cost. Pulse marking differs in detail, since Praat refines pulses by cross-correlation and this marks the waveform's peaks, so expect agreement to a few hundredths of a percent rather than to the last digit.

Explore Our Tools

Browse all tools

Formant Analyzer

Track the formants of a speech recording by linear prediction, Praat's method: F1 to F5 with their bandwidths every 10 ms, pitch and voicing on each frame, a vowel chart of F1 against F2 with reference vowels, and a CSV or JSON table for the notebook.

Pitch Detector

Track the fundamental frequency of a recording over time with the YIN difference function, and read the median pitch as a note name, a MIDI number, and its deviation in cents.

Audio Quality Analyzer

Read an audio file's spectral cutoff, clipped sample runs, DC offset, noise floor, and true peak in one pass, and compare the cutoff against the frequency each encoder usually stops at.

Audio Recorder

Record from a microphone straight to WAV, MP3, M4A, Ogg, or FLAC, with a level meter before you start and a choice of sample rate, channels, and bit depth. Echo cancellation, noise suppression, and automatic gain are all off.

Spectrogram

Plot how an audio file's spectrum changes over time, with an FFT size from 512 to 8192, a log or linear frequency axis, an adjustable floor, and a PNG export.

Audio Feature Extractor

Measure an audio file frame by frame: RMS, zero-crossing rate, spectral centroid, spread, rolloff, flatness, flux and 13 MFCCs, as CSV or JSON with a time column, for notebooks, classifiers and plots.

Dynamic Range Meter (DR Value)

Measure the DR value of an audio file the way the TT dynamic range meter does: the loudest fifth of its 3-second blocks against the second-highest sample peak, per channel and as the integer the loudness-war database files a release under, next to its LUFS and loudness range.

Reverb Time Analyzer (RT60)

Measure a room's reverberation time from a recorded clap, balloon pop or impulse response: T20, T30 and EDT per octave band from 63 Hz to 8 kHz by backward integration, the ISO 3382 method, with the decay curve, the fit and how far each band sits above the noise floor.

Audio Metadata Viewer

Read the container, codec, sample rate, channel count, duration, average bitrate, and embedded title, artist, and album tags of an audio file, and copy the whole report as JSON.

Decibel Meter

Measure sound level from the microphone with A, C, or Z weighting, Fast, Slow, or Impulse response, a running Leq, Lmax, Lmin, L10 and L90, and a calibration offset that turns the digital reading into a sound pressure level.

Impulse Response Capture

Measure a room's impulse response with a sine sweep: a sweep plays through the speakers, the microphone records the answer, the recording is deconvolved to the room's response and saved as a WAV, with its decay, frequency response and mid-frequency reverberation time read on the spot.

Key Detector

Find the musical key of an audio file from its chromagram, with the Camelot code, the confidence, the relative and parallel alternatives, and the full 24-key score table.