Voice analyzer: what it measures, healthy ranges and free tools

A voice analyzer is software that measures your voice from a recording, usually pitch, volume, pace, pauses, filler words and voice quality. The numbers are useful, but only if you know what each one means and which ones are too fragile to trust. This guide gives you both, including the free tools we could verify.

What does a voice analyzer measure?

A voice analyzer turns sound into numbers across three groups: acoustics such as pitch and volume, delivery such as pace and pauses, and voice-quality values such as jitter and shimmer. Basic tools only show live pitch and volume, while speech coaching apps add word-level measures from a transcript.

Here is what the main measures mean, along with reference points we could trace to a source. Treat them as orientation. Many have no official "healthy" number at all, and we say so where that is the case.

Sources: pitch from Berg et al., 2017 and de Felippe et al.; pauses from Campione and Véronis, 2002; volume from NIDCD; jitter and shimmer thresholds from the Praat manual pages on jitter and shimmer; the higher HNR figure from a study of Iranian speakers. For more on pitch, see average voice pitch, and for pace, see average speaking rate.

Why do voice analyzers disagree with each other?

Voice analyzers disagree because they use different algorithms, microphones and rooms. For pitch the differences are usually small. For fine measures such as jitter they can be enormous, so a value only means something against earlier readings from the same tool in the same conditions.

The Praat manual gives a striking example. For a constant-pitch test sound with a little added noise, Praat reports a jitter of 0.02% and the MDVP program reports 0.6%, because they locate voice cycles in different ways (Praat manual). That is a thirtyfold difference on the same sound, and neither program is broken.

Your phone is not the main problem. A 2023 study found that modern smartphones can capture recordings good enough for informative acoustic analysis, although device, setting and microphone distance all had measurable effects (Awan et al., 2023, Journal of Voice).

How to get a useful reading

The way to get a useful reading is to control the setup so that any change you see comes from your voice. Use the same quiet room, phone position, passage and time of day. Then take several readings and watch the trend across weeks rather than reacting to any single number.

1. Choose your room (1 minute). Pick somewhere soft and quiet, such as a bedroom. Turn off fans and step away from hard walls.
2. Fix the distance. Hold the phone about 15 to 30 cm from your mouth, slightly to one side, and use that distance every time.
3. Warm up gently (1 minute). Hum, and say a few sentences. Your first words of the day read lower and rougher.
4. Read the same passage (30 seconds). Reading gives you comparable pitch and clarity numbers. A reading aloud practice passage works well.
5. Speak freely (30 seconds). Pauses and fillers only mean much in unscripted speech, because a script hides them.
6. Repeat three times and take the middle result. Do this once a week and write the numbers down. Daily readings mostly capture noise.

What are the best free voice analysis tools?

The best free option depends on your question: quick browser tests for pitch and pace, Pitch Witch for pitch and resonance on iPhone, and Praat if you want research-grade analysis and do not mind a steep learning curve. All of these were checked in November 2026; listings can change.

Our three browser tools run on your device, so your audio never leaves it. Voceris itself is a different kind of tool: you read a short prompt for 15 to 35 seconds and it scores depth, resonance, clarity, stability, pace, pauses, filler words, vocal fry and volume. Those scores are its own training scale, not clinical values. The first voice score (tier and overall score) is free. The full breakdown, plan and drills need Voceris Pro, and it analyzes spoken English only. For a broader comparison, see the best voice training apps.

What can a voice analyzer not tell you?

A voice analyzer cannot diagnose a condition, reveal your personality, detect lies or say how attractive you are. It measures properties of sound. Claims that go further are interpretations, and they should come with evidence you can check, ideally from independent research and not just the app's own marketing.

It is not a medical test. Crossing a threshold does not mean you have a disorder, and passing does not mean you are healthy. Recommended clinical protocols combine acoustic measures with imaging of the vocal folds and aerodynamic tests (Patel et al., 2018, American Journal of Speech-Language Pathology).

It is not a personality reading. Listeners form quick and consistent impressions from a voice, even from the single word "hello," but the researchers themselves framed those judgments as impressions, "accurate or not" (McAleer, Todorov and Belin, 2014, PLOS ONE). A tool can estimate how you might come across. Be skeptical of one that claims to know who you are.

It is not a lie detector. In a US National Institute of Justice field test with 319 arrestees, two commercial voice stress programs caught about 15% of those who lied about recent drug use and had roughly 50% overall accuracy, or about the same as flipping a coin (Damphousse, 2008, NIJ Journal). Forensic voice comparison is a separate, specialist field that consumer apps do not perform.

It is not a gender verdict. Tools that place you on a male-to-female scale mostly reflect pitch, and sometimes resonance. Those measures overlap between people and cover only part of how a voice is perceived. If you are training for gender-affirming reasons, a specialist speech-language pathologist gives feedback an app cannot.

It is not an attractiveness score. Some apps put a single number on this. Attraction depends on the listener and the context, so any such number is the app's own opinion in numeric form.

What should you do with your results?

Pick one measure that matters for your goal and train it for two to four weeks, then measure again under the same conditions. Most people gain more from fixing one audible habit than from studying nine charts, and they can usually hear the difference themselves.

When to see a professional

See an ENT doctor or speech-language pathologist if you have hoarseness lasting more than two to three weeks, pain when speaking or swallowing, frequent voice loss or a sudden voice change. Do this even if every number in your app looks fine, and do not treat a bad score as a diagnosis. NIDCD's voice care page is a good starting point.