Ever had a song stuck in your head, but haven’t been able to identify it? Of course you have. It could be a hook or a lyric snippet you’re thinking about, but you just can’t name the song or the artist. This common predicament can lead to hours or even days of frustration, with the answer always just out of reach as the cogs in your brain whirr and grind.
Well, tech is here to help: The song recognition tools that have been developed over recent years are able to match up patterns of music and lyrics. You might be surprised at the quality of the results that turn up with very little information.
A lot of the tools we’re covering here do dual duty—they can recognize songs that you hum or sing part of, and they can also name tracks if the official recordings are being played within range. We’re focusing on the first job here, but you might find the other music recognition features useful too.
Google Assistant and Siri
Google Assistant will try its best to identify a song for you. Credit: David Nield
You can load up your favorite digital assistant on your phone or tablet and have it try and identify a song for you. Just load up Google Assistant or Siri, ask “What’s this song?” and try singing or humming a part of it, or saying some of the lyrics.
Google Assistant or Siri will then do their best to find a match. In our testing, Google Assistant is more adept at identifying songs from fragments, whereas Siri tends to need a lyric line or two. Of course part of the success rate is going to be down to the quality of your singing or humming.
Both these digital assistants do well at music identification when you’ve actually got the song itself playing—though most of the time that you’re playing a track, you’ll know what it is, unless it’s in the background of a film or show, or it’s being played in public.
Shazam and SoundHound
We’ve managed to get SoundHound to recognize tracks from our humming. Credit: David Nield
These are the two leading apps when it comes to identifying music: You can pick up Shazam for Android and iOS, and SoundHound for Android and iOS. As you probably know, these apps are adept at identifying songs playing around you, but they’re also able to pull up artist and track name info from audio fragments.
Both apps put music discovery front and center, and all you need to do is tap the big Shazam or SoundHound button on the app’s main screen to get it to start listening. You can try humming, whistling, singing, or reciting lyrics to get a match.
In our tests, SoundHound did a little better than Shazam on recognizing the songs we were trying to do a passable impersonation of—though there are a lot of factors at play here, from the uniqueness of the track, to how accurately you’re humming or whistling it.
Google and YouTube
YouTube is also able to recognize a song from a hummed melody. Credit: David Nield
If you can remember a few of the lyrics of the song you’re trying to identify, then typing them into Google (or your favorite search engine of choice) can often be enough to turn up the name of the track, assuming the lyrics aren’t too generic.
Approximations are okay too—we were recently able to use a Google search to identify a song with the lyrics “hustle to be free” even though what we’d typed in was “hustle to the sea.” Just enter as much as you can remember and see what comes up.
Another Google-owned app, YouTube, has the humming recognition powers that Shazam and SoundHound do. However, it’s only in limited testing at the moment, so it might not be available for you: In YouTube for Android, tap the search icon (the magnifying glass, top right), then the mic icon (top right), then Song.
ChatGPT and Copilot
Give ChatGPT some details, and see what it comes up with. Credit: David Nield
Generative AI is pretty much everywhere nowadays, and it can be handy for identifying songs you’ve got stuck in your head. Tools like ChatGPT from OpenAI and Copilot from Microsoft can’t understand humming or whistling, but they can understand and combine lots of other bits of information.
For example, if you remember half a lyric, the style of the song, and the nationality of the singer, then there’s a good chance an AI tool will be able to give you an answer. Any kind of clue will help, from the era it was first released in, to where you might have heard it (on the radio or on the soundtrack to a film, for example).
We’ve managed to get both ChatGPT and Copilot to identify some pretty obscure songs based on details like what’s in the accompanying music video, and how many band members there are. However, you won’t always get the right answer—remember that these AI bots are far from infallible.
Source : Popular Science