How we can help

Need help with your Managed IT Services?

Our team are available Mon – Fri: 7:30am-5:30pm

Call Now On:
Stourport: 01299 848311 Hereford: 01432 663026

Technical Support

Contact us

- 22nd Oct 2020

‘Hum to Search’ Songfinder

Google has introduced a new feature to its search that enables it to identify a song that a user hums or whistles.

What’s This Song?

Serving the same basic purpose a the Shazam app, for example, users can now ask the Google app, Google Search widget (by tapping the mic icon) or Google Assistant the question “Hey Google, what’s this song?” followed by 10-15 seconds of humming or whistling the melody.

The way that Google’s service adds value in relation to simply identifying a song (like Shazam) is that it only needs the input as a hum rather than actually being played the original tune.  Also, it adds value by delivering what it judges to be the strongest (percentage) song matches, and the user can then select the best match, explore more information about the song and artist, view music videos by the artist, find lyrics and or listen to the song on their favourite music app. Users are also given information about other recordings of the song. 

20 Languages on Android

Although the new feature only works in English on iOS, it is available in 20 languages on Android.  Google says that it hopes to expand this to even more languages in future.

Builds On Previous AI Research

Google says that the new feature builds on the work of their AI Research team’s music recognition technology that was launched on the Pixel 2 in 2017 that uses deep neural networks to bring low-power recognition of music to mobile devices. In 2018 this same technology was used in the Google app’s SoundSearch feature which was linked to a catalogue of millions of songs.

New

This latest version is able to identify melodies that are hummed or whistled because it uses machine learning models that have been trained to identify songs based on a variety of sources, including humans singing, whistling or humming, as well as studio recordings.  These models transform the audio into a number-based sequence representing the song’s melody.

Algorithms are used to take away any other details, such as accompanying instruments and characteristics of a voice thereby simply leaving a song’s number-based sequence – a kind of song fingerprint.  It is this sequence that the software is able to recognise when a person simply hums a tune.

What Does This Mean For Your Business?

For Google, this is another way that it is expanding its use of machine learning and AI to make features that are clever and engaging, thereby helping Google compete for how much time and attention we give its platform compared to competitors like the social media giants and in its traditional fight against the other search engines.  Google has been adding more of these engaging features in recent times, such as a new spelling algorithm for search queries,  using AI to enable in-video searches, introducing data sources to Google search as well as new ways to use its Lens and augmented reality (AR) features.  Outside of search, Google has also been competing for our attention and time by agreeing to pay publishers to create and curate news for its ‘Google News Showcase’, updating its Android OS and releasing an updated Nest Audio smart speaker.  Businesses of all kinds use google search, both to be found in and for research so the addition of many new features using new technologies will make a positive contribution to this.  The ‘humming a song’ feature is, of course, likely to benefit music artists and publishers but may also be a way for Google, in the wake of the running down of Google Play music, to further promote its YouTube service.

Google Rating
5.0
Based on 45 reviews