How the Voice Gender Detector Works
A voice gender detector measures two independent acoustic properties and combines them rather than relying on pitch alone. Fundamental frequency (F0) is how fast your vocal folds open and close, and it is what you perceive as pitch. Averaged across a sentence, typical adult masculine voices centre roughly around 85 to 155 Hz and typical feminine voices roughly around 165 to 255 Hz: population averages with wide individual spread and a real overlap between them. Formants are the resonant frequencies of your vocal tract; the tool extracts the first two, F1 and F2, using LPC spectral envelope analysis. The two lines of evidence are combined with pitch weighted at 68% and formants at 32%, because pitch alone systematically misreads two common cases: a masculine voice in falsetto has a high F0 but retains long-vocal-tract resonance, and a naturally low feminine voice has the opposite pattern. On short, connected speech the formant cue is weak, so it only partly corrects those two cases. That is why the result page also offers an optional deep resonance check, described below.
Formants sit in the throat and mouth cavity above the vocal folds. Their positions depend mainly on the length of that tract, which is on average shorter in feminine anatomy, pushing resonances higher. F1 and F2 are read from an LPC spectral envelope computed on a downsampled copy of your signal.
Reading Your Result
You get three numbers rather than a single label, because a single label would hide how the conclusion was reached.
Example result · sample recording, not your voice
The androgynous band is not a hedge. Human voices do not fall into two separated clusters; the distributions overlap in the middle, and voices in that region are accurately described as sitting between the two poles. The two sections below walk through the raw measurements panel, including the F1 and F2 reference points, and how the confidence figure is calculated.
Is My Voice Feminine or Masculine?
If you arrived here looking for an "is my voice feminine or masculine" test, this page is that test. Press record, speak one relaxed sentence, and the detector returns a percent-feminine figure, a label, and the pitch and formant readings the label was built from. Whether you phrase the question feminine-first or masculine-first, it is the same measurement of the same two acoustic cues, so the answer does not change with the wording.
A gender voice detector is not listening for a man or a woman. It measures where your fundamental frequency and vocal tract resonances sit relative to two population distributions, then reports that position. That is why the result is a percentage first and a label second: most voices sit clearly on one side, a sizeable minority sit in the overlap, and a percentage describes both cases honestly where a two-way label cannot.
Voice gender analysis of this kind is most useful when you repeat it. One take tells you where you are today; three takes over a week, recorded under the same conditions, tell you whether the number is stable or moving. If you are working on your voice, the percent-feminine figure and the F1 and F2 readings are the ones to log, since they move before the label does.
Does My Voice Sound Male or Female?
Questions like "does my voice sound feminine or masculine" and "what gender do I sound like" are about perception: what a stranger on the phone would probably assume. The detector answers with the two cues listeners lean on most, speaking pitch and vocal tract resonance. Pitch carries the larger share, so it is weighted at 68% and resonance at 32%.
Some people search for a voice gender identifier, or for a tool that can guess gender by voice. This page does not identify anyone. It makes no claim about who is speaking and keeps nothing, because the recording is analysed in your browser and never uploaded. What it reports is where the sound itself sits: 68% feminine or higher reads Feminine, 32% or lower reads Masculine, and everything in between reads Androgynous.
A boy or girl voice detector is a different job. Before puberty, children of every gender tend to speak well above the 158 Hz crossover this scale is built around, so a child's recording will usually land on the feminine side whatever the child's gender. For young voices, read the result as a description of a high voice. It is not an answer to boy or girl.
To learn how you actually come across, record the way you talk to people every day. A careful reading voice is not the one others hear. Take three recordings and keep the middle result, as described under Getting a Reliable Reading below.
Feminine Voice Test and Masculine Voice Test
Most people who search for a feminine voice test or a masculine voice test are not asking which box they fall in. They want to know how far they are from the side they are aiming for, and what would move the reading. Because the scale is a fixed formula rather than a trained black box, those questions have exact answers.
Feminine voice test: where the Feminine label starts
The detector labels a voice Feminine at 68% or higher. On pitch alone, that line sits at a median speaking pitch of about 170 Hz. When resonance is measured as well, it carries 32% of the weight, so it moves the line: with a neutral resonance reading the Feminine label starts at about 177 Hz, a moderately feminine resonance pulls it down to about 171 Hz, and a moderately masculine resonance pushes it up to about 184 Hz.
That is why raising pitch alone often stalls in the Androgynous band. If your pitch is already above 180 Hz and the result still reads Androgynous, the resonance readings (F1 and F2) are the part to work on, and the deep resonance check below tells you whether the voice as a whole is heard as feminine, not just the pitch number.
Masculine voice test: how masculine is my voice?
The Masculine label starts at 32% feminine or lower. On pitch alone, that is a median speaking pitch of about 146 Hz or below. With a neutral resonance reading the line moves to about 139 Hz; a moderately masculine resonance lifts it to about 145 Hz, and a moderately feminine resonance lowers it to about 132 Hz. Lower F1 and F2 averages, the sound of a longer vocal tract, are what make a voice read masculine at a pitch that would otherwise sit in the middle.
For a sense of scale on the pitch cue by itself: 140 Hz reads 25% feminine, 158 Hz sits exactly halfway at 50%, and 180 Hz reads 80% feminine. A few hertz around the middle moves the pitch cue a long way, which is also why one tired or excited take can flip the label. Record three takes and read the middle one.
The Raw Measurements Panel Explained
Open the raw measurements panel under your result and the analysis exposes its inputs. These are the values worth writing down if you plan to compare takes over time.
F0, in hertz. Your average speaking pitch across the voiced portions of the take. Typical adult masculine voices centre roughly around 85 to 155 Hz and typical feminine voices roughly around 165 to 255 Hz, with a genuine overlap zone between about 155 and 175 Hz. On this analyzer's scale, the crossover sits near 158 Hz: sentence-averaged pitch above that begins to lean the pitch cue feminine, and below it masculine, gradually rather than as a hard switch.
F1 and F2, in hertz. The first two formants, averaged across the take. Because formants swing widely with each vowel you happen to utter, single-word values mean little: the averaging across a full sentence is what makes them usable. Longer vocal tracts produce lower formants, so lower averages lean the resonance cue masculine; in this analysis, sentence averages above roughly 500 Hz for F1 and roughly 1550 Hz for F2 begin to lean it feminine, with F2 weighted somewhat more heavily than F1.
Signal quality figures. The panel also reports the harmonic-to-noise ratio and the share of the take that was clearly voiced. These do not shift the gender scale directly, but they feed the confidence calculation: a take that was mostly breath, background noise or silence does not deserve a confident verdict, and the analysis says so rather than bluffing.
How the Confidence Figure Is Calculated
Confidence in this analysis measures the reliability of the measurement, not the extremity of the result: a distinction many analyzers get wrong. It is built from three factors multiplied together, and knowing them tells you exactly how to respond to a low score.
Cue agreement is the largest factor. Pitch and resonance are independent lines of evidence, so when both point the same way, including when both point at the middle, the analysis is on solid ground. When they contradict each other, confidence drops sharply, and that disagreement is itself the finding worth acting on.
Evidence completeness comes second. If the formant extraction fails, usually because of a muffled microphone or a very quiet take, the analysis falls back to pitch alone and caps its own confidence, because a single cue systematically misreads falsetto and naturally low feminine voices. Signal quality is the third factor: low voiced ratio or poor harmonic-to-noise ratio scales the whole figure down. So a low confidence score is a diagnosis, not a shrug: re-record closer to the microphone in a quieter room, and if it stays low with good signal, the cue disagreement is real.
Male or Female Voice Test: How Accurate Is It?
People search for a male or female voice detector, a voice checker for male or female voices, or a masculine or feminine test, and all of them land on the same acoustic question. This tool uses the words masculine and feminine deliberately, because it measures how a voice sounds rather than who is speaking. Read those labels as male-typical and female-typical resonance and the result means exactly what the search phrasing intends.
For voices well away from the middle, accuracy is high, because the pitch gap and the formant gap between typical masculine and feminine speech are both large and both point the same way. Errors cluster in two situations. Near the 155 to 175 Hz overlap the two populations genuinely cross, so any acoustic method, including a trained listener, becomes less certain; and a muffled or noisy take can lose the formant cue entirely, leaving pitch alone to decide, which misreads falsetto and naturally low feminine voices.
The confidence figure exists so you can see which situation you are in. A confident verdict with both cues agreeing is as reliable as a short recording can be. A low-confidence verdict is not the tool failing: it is telling you to re-record closer to the microphone in a quieter room, and if it stays low, that your voice sits where the categories overlap.
The Androgynous Band: When the Test Says Neither
Roughly one voice in six lands in the middle band rather than at either pole, and that is the result people most often assume is a malfunction. It is not. Between about 155 and 175 Hz the masculine and feminine distributions genuinely overlap, so an androgynous voice test result means the measurement worked and your voice sits in the region where the two populations cross.
Use the percentage, not the label. If you are asking how feminine your voice is rather than which of two boxes it falls in, the percent-feminine figure is the answer and the label is the blunt instrument. A voice moving from 38% to 47% has changed measurably while its label has not moved at all. This is why the number is shown first and the label second.
The middle is a position, not an error. A masculine-or-feminine voice test can only report where the acoustics sit relative to population averages. Landing between them is a description of resonance, not an unresolved question about you.
Using It for Voice Training
Many people arrive at a tool like this while working on their speaking voice, whether for gender-affirming voice training, acting, or professional presentation. Used carefully it is genuinely helpful, and there are three things worth knowing.
Watch resonance, not just pitch. Raising pitch alone is the most common early approach and it produces the least convincing result, because listeners respond strongly to vocal tract resonance. In the raw measurements panel, F1 and F2 are the resonance figures. Pitch usually shifts first during training while formants follow more slowly, so a rising percent-feminine score with flat formants tells you where to direct the next stage of practice. The deep resonance check gives a second opinion that a raised pitch alone cannot sway.
Keep conditions identical. Same room, same device, same distance, same passage read aloud, ideally the same time of day. Formant estimates in particular are sensitive to microphone placement, and a change of setup can produce an apparent shift that is entirely artefact. The guide on tracking voice training progress shows how much a reading moves between takes on its own.
Never push into strain. No number on any website is worth an injured voice. Pitch reached by squeezing rather than by adjusting resonance shows up here as rising jitter and falling harmonic-to-noise ratio, so if your gender reading improves while those two deteriorate, that is a warning sign worth respecting. Working with a speech-language pathologist or a qualified voice teacher is the safe path; this tool is a mirror between sessions, not a substitute for one.
Deep Resonance Check: Is It Just My Pitch?
The standard reading leans heavily on pitch. We tested how much: we took clean recordings of 20 men and 20 women from the public LibriSpeech corpus, then used Praat to raise only the pitch of each man to about 200 Hz and lower only the pitch of each woman to about 120 Hz, leaving resonance untouched. The standard reading moved with the pitch every time, labelling all 20 pitch-raised men feminine and all 20 pitch-lowered women masculine. The full method and data are in our pitch vs resonance guide. Listening research points the other way: in Hillenbrand and Clark's 2009 study, changing the perceived sex of a sentence usually took a shift in both pitch and formants.
So after a recording on this page you can run an optional deep resonance check. It loads an open-source speech model (Common Voice Gender Detection, Apache-2.0) into your browser and asks it whether the take sounds like a typical female or male voice. On the same 40 speakers, cut to 5-second takes, it labelled 37 of the 40 original recordings correctly, and it was fooled by only 3 of the 20 pitch-raised men and 2 of the 20 pitch-lowered women.
The check then sets its answer beside your pitch cue. If your pitch reads feminine but the model still hears a masculine voice, resonance has probably not moved with your pitch yet, which is the most common plateau in voice training. The first run downloads about 95 MB of model files from Hugging Face. The recording itself is analysed on your device and is never uploaded. The model was trained on recordings labelled female or male, so it describes how a voice tends to be heard, not who anyone is, and it can be wrong on short or noisy takes.
Getting a Reliable Reading
This is a description of sound, not a statement about identity. Acoustic measurements describe how a voice resonates, and nothing more. A voice reading as masculine or feminine on an acoustic scale says nothing whatsoever about who its owner is.
The underlying reference ranges also come from population averages that vary by language, accent, age and individual anatomy. Children's voices sit high regardless of gender. Age shifts everything: feminine voices tend to lower gradually through adulthood while masculine voices dip and then rise again later in life. And a heavy cold can move both pitch and formants enough to change a borderline result.
Finally, a single five-second take is a snapshot. Read a sentence in your ordinary relaxed voice rather than performing one, record two or three times, and treat the middle result as the representative one.
If you want the full picture instead of just the gender read-out, the main voice rater gives you a 1-10 score and a voice age estimate from the same five-second recording.
Frequently Asked Questions
Last reviewed: 2 October 2026
Is my voice masculine or feminine?
Record five seconds and you get a direct answer with the evidence behind it. The tool measures your fundamental frequency and your formant resonances, then places your voice on a masculine-to-feminine scale with a confidence figure alongside. People often phrase this as whether their voice sounds male or female, which acoustically is the same measurement. Whichever label comes back, the raw numbers are shown next to it so you can judge the result yourself rather than take it on trust.
Can raising my pitch fool a voice gender test?
It can fool the standard reading on this page, which leans heavily on pitch: in our test of 40 public recordings, raising only the pitch of 20 men to about 200 Hz made the standard reading call all 20 feminine. The optional deep resonance check, which runs a speech model in your browser, was fooled by only 3 of those 20. If the two disagree, your resonance has probably not caught up with your pitch yet.
Male or female voice test: how accurate is it?
Reliable for voices clearly on one side of the scale, where pitch and formant differences are large and consistent, and less certain in the 155 to 175 Hz overlap where the two populations genuinely cross. The confidence figure tells you which case applies to your take. If it reads low, re-record closer to the microphone in a quiet room before reading anything into the label.
Is this a male or female voice detector?
Yes, in the acoustic sense: it measures the same pitch and resonance cues that make a voice sound male-typical or female-typical to a listener. The result uses the words masculine and feminine because the tool describes sound, not the person speaking. Search for it as a voice checker for male or female voices or as a gender voice detector and you get the same measurement.
What does a voice gender analysis measure?
Two things: fundamental frequency, the rate your vocal folds vibrate, and the first two formants, the resonances of your throat and mouth. Both are extracted from a five-second recording on your own device and combined with pitch at 68% and resonance at 32%. The raw values are shown in the measurements panel so you can check the analysis yourself.
How does a voice gender detector work?
Two acoustic properties do the work. The first is fundamental frequency, the rate your vocal folds vibrate, which differs on average between typical masculine and feminine voices. The second is formant frequencies, the resonances of your vocal tract, which depend on its length. Both are measured on your device and combined, with pitch weighted at 68% and resonance at 32%.
Why does it say androgynous?
Because your voice genuinely sits in the overlap region, roughly 155 to 175 Hz, where masculine and feminine voice distributions cross. That is a real acoustic finding rather than the tool failing. The confidence figure shown alongside tells you how close to the boundary you are.
Is a voice gender detector accurate?
For voices well away from the middle it is reliable, because pitch and formant differences there are large and consistent. Near the boundary no acoustic method can be definitive, since the underlying distributions genuinely overlap. It describes how a voice sounds acoustically, which is not the same thing as identity.
Can I use this for voice training?
Yes, and it is one of the more useful applications, provided you track resonance and not only pitch. Keep your recording conditions identical between sessions and watch the formant figures in the raw measurements panel alongside the pitch figure, since resonance often lags behind pitch during training.
How feminine is my voice?
The percent-feminine figure answers that directly: it places your voice on a continuous scale rather than sorting it into one of two boxes. Watch that number rather than the label if you are tracking change, because it moves long before the label does. The same figure read from the other direction tells you how masculine your voice is; they are two ways of describing one position on one scale.
Is this voice gender analyzer free, and does it run online?
Yes to both. It runs online in your browser with nothing to install, no account and no payment. Because the analysis happens on your own device, your recording is never transmitted anywhere, which is also why it works on a phone without an app.
What pitch counts as a feminine voice?
Typical adult feminine speaking voices centre around 165 to 255 Hz and typical masculine ones around 85 to 155 Hz, with substantial individual variation and a genuine overlap zone between them. Pitch alone is not decisive, which is exactly why formants are measured too.
How masculine is my voice?
Record one relaxed sentence and read the percent-feminine figure from the other side: the lower it is, the more masculine your voice reads. The Masculine label starts at 32% or lower, which on pitch alone means a median speaking pitch of about 146 Hz or below. Lower F1 and F2 resonance readings push a voice further toward masculine at the same pitch.
What pitch counts as a masculine voice?
Typical adult masculine speaking voices centre around 85 to 155 Hz. In this detector, pitch alone reads Masculine at about 146 Hz or below and Feminine at about 170 Hz or above, and a moderately masculine or feminine resonance reading moves both lines by roughly 6 to 7 Hz.
What is a normal F1 and F2 for a voice?
Formant values vary a great deal with the vowels you happen to speak, which is why the analyzer averages across a full sentence. Broadly, longer vocal tracts produce lower formants, so masculine-typical voices tend to show lower F1 and F2 averages than feminine-typical ones. In this analysis, sentence-averaged values above roughly 500 Hz for F1 and roughly 1550 Hz for F2 begin to lean the resonance cue feminine.
Why is my confidence score low?
Three things can pull it down: the pitch cue and the resonance cue disagreeing with each other, formants failing to extract so only pitch is available, or a noisy take with little clearly voiced sound. A low confidence with disagreeing cues is the most informative case, because it usually means pitch and resonance have moved by different amounts.
What gender do I sound like?
Record one relaxed sentence and the result shows a percent-feminine figure with a label: Feminine at 68% or above, Masculine at 32% or below, and Androgynous in between. The figure comes from your speaking pitch and your resonance, the two cues listeners rely on most, so it is a reasonable stand-in for how a stranger would hear you. Record three takes and keep the middle one.
Can it tell whether a child is a boy or a girl?
No. Children of every gender usually speak well above the 158 Hz crossover this scale uses, so a young voice tends to read feminine whatever the child's gender. The tool describes how a voice sounds and is not a boy or girl voice detector. For adults it compares your voice with adult population ranges, which is what the scale was built for.
Is my recording uploaded?
No. Everything is analysed locally in your browser. No audio is transmitted, stored or retained anywhere.