What Can You Listen With Your Eyes Actually Does

Can You Listen With Your Eyes is a web-based tool that lets you identify songs by humming, whistling, or singing a melody into your microphone. It was originally developed by the research lab at Queen Mary University of London. The system takes your audio input, extracts the melodic contour, and searches a database of music for matching patterns. It works best when you can produce a relatively clear hum or whistle. Background noise, poor microphone quality, or singing off-key will all reduce accuracy. I spent several months evaluating this tool during a project where I needed to identify obscure film scores and jazz improvisations that didn't show up cleanly on commercial platforms. Most of what I learned came from repeated failures rather than instant successes.

Can You Listen With Your Eyes How It Actually Works

You open the website, grant microphone access, and start producing sound. The system converts your audio into a spectrogram representation, then runs pattern matching against a database that includes sheet music transcriptions and commercial recordings. The underlying technology uses something called audio hashing combined with melodic similarity algorithms. That means it isn't looking for exact waveform matches. It's looking for shape matches in the frequency domain. The result is a ranked list of possible matches, each with a confidence score. You usually get three to five strong candidates. From my testing, about 60 to 70 percent of well-performed queries return the correct answer on the first try. That drops to maybe 40 percent if you're working with a fragment that only lasts four to six seconds. Here is a practical problem I ran into. I was trying to identify a piece from a 1970s French film soundtrack. The melody was complex, modal, and kept shifting tonality. The system kept returning results for completely different pieces that shared similar contour shapes but nothing musically related. I got around it by recording a much slower, more deliberate hum at a single steady pitch. Slowing the tempo forced the melody into clearer pitch boundaries. The system matched it correctly on the third attempt after that adjustment.
Download or Access Options The primary way to use this tool is through the official website at query.nls.uk or the current hosted version at canyoulistenwithyoureyes.com. There is no standalone desktop application. You can also access a research API version if you are building something larger, though that requires signing up for academic or development access through Queen Mary's systems.

When This Tool Fails and What to Do Instead

This system has real limitations. It struggles with rhythmic complexity because the algorithm prioritizes pitch contour over rhythm. It struggles with polyphonic music because it assumes a single melodic line. It struggles with pieces that are heavily ornamented or vary significantly between performances. A Baroque fugue, for instance, will almost never come back correctly because the same melodic material appears in multiple voices and the system picks the wrong one. If you are dealing with lyrics or a singable melody from a pop song, Shazam or SoundHound will give you better results in most cases. If you are working with classical music or obscure recordings, this tool still has value even though its database is smaller than commercial alternatives. The setup process is straightforward. You need a browser with microphone permissions, a reasonably quiet environment, and a source melody you can reproduce vocally or instrumentally. No account is required for basic use. Results typically load within five to fifteen seconds depending on how large your query segment is and how many database entries it needs to compare against.