A spectrogram viewer shows how the frequencies in a sound change over time. Use your microphone or open an audio file, and this free online spectrogram draws time from left to right, frequency from bottom to top and loudness as colour.
Spectrogram Viewer
Move over the spectrogram to read time, frequency, note and level.
Nothing is recorded or uploaded: sound is analysed in your browser.
Related Tools
- Frequency Detector
Measure one frequency in Hz, with a live spectrum analyzer.
- Audio File Pitch Detector
Find the notes and pitch curve of a recording.
- Frequency to Note Converter
Turn a frequency you read here into a note.
It is useful for examining vocals, instruments, harmonics, background noise, resonances and transients that are hard to see in a waveform. For a full walkthrough of the format, read how to read a spectrogram.
How to Use the Spectrogram Viewer
- Press Microphone for a live, scrolling spectrogram of the last 10 seconds, or Open audio file to see a whole recording (MP3, WAV, M4A, OGG or FLAC, up to 50 MB and 10 minutes).
- Move the mouse over the picture, or tap it on a phone, to read the time, frequency, musical note and level in dB at that point.
- For files, press play: a white line follows the audio. Click anywhere on the picture to play from that moment.
- Adjust the view: Scale (log, linear or mel), Colours, FFT size, Top frequency and Contrast. Pause the live view to study it.
- Press Save image (PNG) to download the spectrogram with its axes.
A good starting point is log scale, FFT 4096, top frequency 8 kHz and medium contrast. For recordings with a lot of background sound, see noise and background interference.
How to Read a Spectrogram
- Horizontal axis: time. Left to right shows how the sound develops.
- Vertical axis: frequency. Low sounds are near the bottom, high sounds near the top.
- Colour: level. Brighter colours mean more energy at that time and frequency.
A sustained note makes horizontal lines, a drum hit makes a short vertical burst across many frequencies, and noise such as a fan or “sh” makes a wide cloud. The cursor readout turns any point into exact numbers, so you do not have to guess from the colours.
Spectrogram vs Spectrum Analyzer
A spectrum analyzer shows how much energy each frequency has at one moment. A spectrogram adds time: it shows when a 440 Hz tone starts, how long it lasts and what else sounds with it. That is why spectrograms are used for speech, singing, birdsong and music. For a single frequency reading with a live spectrum, use the frequency detector. More background: the difference between pitch and frequency.
Fundamental Frequency and Harmonics
Most musical sounds contain a fundamental, which matches the note you hear, and harmonics at whole-number multiples above it. A note at 220 Hz also shows lines near 440, 660 and 880 Hz, stacked like a ladder. The balance of those lines is the timbre that makes a guitar sound different from a voice on the same note. See why tuners show multiple notes for how this affects tuners.
Spectrogram Decoder: Hidden Images and Messages
Some songs, games and puzzles hide pictures or text in the spectrum: the image is only visible on a spectrogram. To decode one, open the file and set Scale: Linear, Top frequency: Full, Colours: Greyscale, then raise the Contrast. Hidden images usually sit in the upper frequencies, above about 10 kHz, and can take up the whole song or a few seconds. Save the result as a PNG to zoom in. Lossy MP3s may cut everything above 16 kHz, so use the original WAV or FLAC when you can.
What FFT Size Means
The spectrogram is built with the Fast Fourier Transform. The FFT size trades time detail for frequency detail:
| FFT size | Best for |
|---|---|
| 1024 | Fast changes: drums, consonants, clicks |
| 2048 | General viewing, speech |
| 4096 | Voices and melodies: clear harmonic lines |
| 8192 | Fine pitch detail and low notes, slower timing |
A technical explanation is in how FFT works in pitch detection.
Log, Linear or Mel Frequency Scale
On a linear scale, equal distances are equal numbers of hertz, which suits high frequencies and hidden images. A log scale gives every octave the same height (110 to 220 Hz looks as tall as 440 to 880 Hz), which suits music. The mel scale follows how we hear pitch and is the standard view for speech and machine learning. More on octaves: frequency vs note vs octave.
Using a Spectrogram for Voice and Music
- Voice: fundamental and harmonics, pitch slides, vibrato waves, breath noise and consonants. Typical ranges are in frequency ranges for instruments and voices.
- Mixing: low rumble, resonances, harsh sibilance, cymbal energy and instruments fighting for the same range.
- Audio files: check whether an MP3 is a true high-quality file (a hard cut at 16 kHz means it was encoded at a low bitrate).
- Nature and science: birdsong, bat calls in high-sample-rate recordings, machine noise.
For audio work see pitch detection tools for audio engineers, and for the notes of a melody use the audio file pitch detector.
Maximum Frequency and Sample Rate
The highest frequency a recording can contain is half its sample rate: 22.05 kHz for 44.1 kHz audio and 24 kHz for 48 kHz. Choose a lower Top frequency to zoom in on voices or bass.
Microphone, Accuracy and Privacy
Live results depend on the microphone, the room and background sound; phone microphones are weak below about 50 Hz and above about 15 kHz, and automatic processing can change the picture. See how microphone quality affects accuracy. Colours show relative level, not perceived loudness. Files and microphone sound are analysed in your browser: nothing is recorded or uploaded.
Common Questions
What is a spectrogram?
A picture of sound: frequency over time, with colour for loudness.
Why do I see several horizontal lines when I sing one note?
They are the fundamental and its harmonics. The lowest line is usually the note you sing.
Can it show musical notes?
Yes: move over a line and the readout gives its frequency and nearest note with cents.
How do I find a hidden image in a song?
Open the file, choose linear scale, full range and greyscale, and raise the contrast. Look in the high frequencies.
Can I save the spectrogram?
Yes. Save image (PNG) downloads the picture with its time and frequency axes.