# Audio Spectrum Analyzer Load an audio file and press play. The analyzer draws 72 frequency bars across 20 Hz to 20 kHz, names the loudest frequency, reads its level in decibels, and can hold a peak or freeze the display. > Canonical: https://aimusicis.com/audio-spectrum-analyzer ## An audio spectrum analyzer that runs in the page A spectrum view answers a question a waveform cannot: which frequencies are actually carrying the sound. This analyzer keeps a file playing while you change the analysis window, the smoothing and the decibel floor underneath it, so the same passage can be read at different resolutions. ### 72 bars across the audible range The bars are spaced logarithmically from 20 Hz to 20 kHz, so a bass note occupies as much of the display as a cymbal does. ### Three analysis windows Choose a 2048, 4096 or 8192 point FFT. A longer window resolves low frequencies more finely and redraws less often; a shorter one reacts faster. ### Peak frequency, level and note The loudest bin is reported in hertz, its level in decibels, and the nearest equal-tempered note name, using A4 at 440 Hz as the reference. ## Read a spectrum in three steps ### Load an audio file Drop a file or choose one from your device. The analyzer decodes it in the browser; nothing is uploaded. ### Choose a window and a floor Set the FFT size for resolution, then pick a decibel floor: -60 dB shows the loudest detail, -100 dB brings quiet detail up out of the floor. ### Play and read the display Peak hold keeps the loudest frequency on the display after a transient has passed, and freeze stops the drawing while the audio keeps playing. ## Windows, floors and formats ### 20 Hz to 20000 Hz The analyzed band, drawn as 72 logarithmically spaced bars, which is the range most people can hear. ### 3 FFT sizes 2048, 4096 and 8192 points. The window decides the trade-off between frequency resolution and how quickly the bars follow the audio. ### 3 decibel floors -60 dB, -80 dB and -100 dB set the quietest level the analyzer draws, so a quiet passage can be examined without changing the file. ### 6 audio formats MP3, WAV, OGG, M4A, AAC and FLAC, decoded by the browser, up to 90 MB and 15 minutes per file. ## Who reads a spectrum ### Producers checking a mix See where the low end of a track actually sits, and whether a lead instrument is competing with the vocal in the same band. ### Anyone hunting a hum or a rattle Play the passage that sounds wrong and read the peak: a steady tone at 50 or 60 Hz is mains hum, and a narrow spike names the culprit. ### Musicians learning a part by ear The nearest-note readout names the loudest pitch while the passage plays, which is a starting point for working the rest out. ## Tools related to Audio Spectrum Analyzer ### [Key & BPM Finder](https://aimusicis.com/key-bpm-finder) ### [Chromatic Tuner](https://aimusicis.com/tuner) ### [Frequency Generator](https://aimusicis.com/frequency-generator) ### [Guitar Tuner](https://aimusicis.com/guitar-tuner) ## Audio Spectrum Analyzer FAQ ### What does the Audio Spectrum Analyzer show? The Audio Spectrum Analyzer shows how much energy sits in each frequency band of a file while it plays: 72 bars from 20 Hz to 20 kHz, the loudest frequency in hertz, its level in decibels and the nearest note name. ### Does the Audio Spectrum Analyzer upload my file? No. The Audio Spectrum Analyzer decodes the file in the browser with the Web Audio API and draws it on a canvas in the page. Nothing is sent to a server, and no account or credits are needed. ### Which FFT size should I choose in the Audio Spectrum Analyzer? Choose 8192 points to separate two close low frequencies, and 2048 points when you want the bars to follow fast passages. The Audio Spectrum Analyzer uses 2048 by default, which balances the two. ### What does the decibel floor change in the Audio Spectrum Analyzer? The floor sets the quietest level the Audio Spectrum Analyzer draws. At -60 dB only the loudest detail appears; at -100 dB quiet detail is lifted into view along with any background noise in the recording. ### Why does the peak readout show a note name? The note readout translates the peak frequency into the closest equal-tempered note, using A4 at 440 Hz. It names a pitch, so it is approximate: a frequency between two notes reads as the nearer one. ### Can the Audio Spectrum Analyzer listen to a microphone? Where the browser allows it, the Audio Spectrum Analyzer can analyze live microphone input instead of a file. Microphone access is granted by the browser, so the button reports an error if permission is refused or the page is not served over HTTPS. ### Can I save the spectrum as an image or the audio as a file? No. The Audio Spectrum Analyzer is a measuring view, not an editor: it does not record, alter or export audio, and it does not save the drawing. Use a screenshot if you need the picture. ### What are the limits of the Audio Spectrum Analyzer? The Audio Spectrum Analyzer needs a browser with the Web Audio API, reads one file at a time and only while the page stays open. Its bar display covers 20 Hz to 20 kHz, it does not name instruments or transcribe notes, and its peak note is the nearest equal-tempered name for the loudest single frequency rather than a full pitch analysis. ### How many credits does Audio Spectrum Analyzer use? Audio Spectrum Analyzer runs in the browser and draws zero credits. ## Play a tone and watch it land in one bar Generate a steady tone at a known frequency, then check the analyzer names the same note and shows a single spike. _Last updated August 28, 2026_