hum to search
Find a song by humming, whistling or audio.
About
hum to search is a browser-based music recognition tool for the moment when you remember a tune but not the song. You hum, sing or whistle the melody, or switch to audio recognition when the real recording is available, and the site returns song candidates to check by ear.
Two modes cover two different sources. Hum a melody is for a version you perform yourself: humming with closed lips, singing with words or simple syllables, or whistling. The useful information there is the shape of the melody rather than the key, so a steady, comfortable rendition works better than straining for the original notes. Identify audio is for the real recording, either playing nearby or already saved as a file on the device, recorded through the microphone or uploaded as a clip.
A useful sample is short and clear. The guidance is to aim for roughly 8 to 12 seconds of one continuous phrase such as a chorus, hook or repeated line, because a clip under 6 seconds is too short to send and a long recording full of silence or several songs adds noise. The browser checks basic conditions, such as whether the sample contains audible sound and is long enough, before the search is sent.
Results are presented as candidates rather than a verdict. A result begins with the song title, artist and match information returned for the sample, and album details, artwork, duration or links appear when available; empty fields stay hidden instead of being guessed. A no-match state is a complete result rather than endless loading, and the page suggests a next move such as recording a longer connected phrase or switching modes. A technical failure is reported separately with a retry.
Recording stays private in a specific sense: the site states that the short sample is handled for the recognition request and the original recording is not saved to the product database or an audio-storage bucket. Microphone access begins only after a recording action is pressed, and an audio file can be uploaded instead of recording.
The product starts with one combined trial per person, followed by optional one-time credit packs for further searches.
Ready to launch?
Submit your product and get discovered by builders and creators worldwide.
Launch Now
FounderPlaybooks.
What other founders did to grow.
2722 dispatches from hundreds of founders, pulled from the week's best podcasts.
We made it really simple to post a link to your Rapchat to your Twitter and your Facebook — two ugly square buttons — and that worked. We saw a 10x return on that, and to date that type of flow has driven millions and millions of downloads.
Adding two share buttons drove a 10x return — viral loops beat paid acquisition
With no paid marketing budget in the early years, Seth Miller's team obsessively refined the share-and-return loop. Adding basic social sharing (originally just Twitter and Facebook buttons) produced a 10x multiplier on installs. Later iterations auto-generated shareable video of each track — which became the app's most-used feature because native video performs better in social-feed algorithms. The compounding effect of tiny loop improvements, not any single big bet, fuelled 80% organic growth.
I usually go to a popular app and a similar app within the same niche and I look at the layout of their app and I take the elements from it and adapt it to my app so I don't copy it the exact same... try not to reinvent the wheel just take what's working from their app adapt it a little and change it up.
Steal Onboarding Layouts From Category-Leading Apps Then Twist Them For Your Niche
Ethan designs onboarding by lifting structure from incumbents like Cal AI and Duolingo, since their flows reflect years of A/B testing he could never afford. Adapt rather than copy verbatim, then add a niche twist so the screens feel native to your audience. Stand on the shoulders of teams who've already tested every onboarding pattern.
There's a play for whatever you're stuck on.
Read all 2722 playbooks