The turn of this century witnessed a new form of intelligence -- the augmentation of human idea with the computational powers of the machines. These machines have now have become massive data-driven engines.
Machine Learning models like neural networks ingest tonnes of data and churns out results without getting tired. These algorithms have enabled us to see the hidden correlations, which in turn have been exploited for building better models. From recommending designs for nuclear plants to soundtracks, AI has been leveraged in almost all fields.
Music, one of the most celebrated creative traits amongst humans, also has been touched by the far reaches of algorithmic advancements.
This resulted in the development of personalised song recommender systems to cater to one’s own moods.
One such service, Spotify, which has been topping the charts as one of the world’s topmost music streaming providers. Spotify’s personalisation system got better as it got supplemented with extra information such as a soundtrack’s danceability, energy, valence, and more. But the machine learning team at this Singapore based start-up Musiio argues that the inefficiencies of the present AI assisted services still prevail.
Musiio boasts of having an AI that can ‘listen’ to millions of tracks at once, recognise thousands of features from every single audio track and categorise music with an accuracy greater than 90%. These results have landed big for Musiio as it becomes the first venture capital-backed music AI startup in Southeast Asia.
Why Musiio Can Be A Good Alternative To Spotify
Musiio aims at reducing inefficiencies using AI in a way such that users do not have to sludge through Spotify to find their own flavour of music. Moreover, businesses who use Musiio will have the option of automating this search process.
Musiio leverages AI to enable:
- Efficient audio tagging
- Increased search speed
- Improved catalogue recall accuracy to fetch hidden gems of music.
These functionalities along with many others give Musiio the much-needed edge and this was made possible thanks to the state-of-the-art AI models, which can:
- ‘Listen’ to and tag over 200,000 new tracks per day. Once ‘heard’ our technology can search through music at a rate of 1m+ tracks in two seconds.
- Listen to the audio rather than using a popularity algorithm to match users’ audio profile with music they'll love, getting the best out of the whole catalogue and creating a personalised experience, not an echo chamber.
- Can be custom-trained to work from a cold start.
In addition to these features, tag packages are available or customers can request a custom build to fit their personal database.
Users get an improved search experience with enhanced capabilities that allow for audio-reference searching. Their catalogue recall accuracy has been improved to find unique music that fits with your chosen track with less effort. They have also improved their search speed where users can search for more than one million tracks in just below 2 seconds.
By pasting an audio link, an mp3, or a music track into the search tool, users now no longer need to find the words to describe the track they are looking for. Instead, they are able to use a 'reference track’ or a seed track and can spend time enjoying accurate audio matches rather than being caught in a loop of inaccurate keywords and poor search results.
Here are few results when for few songs uploaded from Youtube into the search tool:
For Eminem’s I’m not afraid soundtrack, the results were as follows:
And Vivaldi’s four seasons(Storm):
For Dmitiri Shostakovich’s Waltz No.2 :
Here the percentage represents AI’s predicted probability about the tag. These predictions are quite accurate as one can see the clear contrast of tags between Vivaldi’s adrenaline pumping Storm composition and the melodious Waltz.
Future Of Musiio
The use of AI to disrupt the music industry has been gaining traction of late. Earlier this year, Google demonstrated how to shred notes into lower dimensions and then perform fundamental techniques like batch normalisation and autoregressive factorisation to create new soundtracks from old ones. Though using the word disruption in domains steered by human creativity is still a hyperbole, one can still sense how AI can be used as an augmentation to creative tasks.
To create and share the work easily; making it accessible to the wider population is what any artist wants. And, AI-driven technologies like those of Musiio’s will keep trying to fill the voids left out by technical impediments in the past.
Find your favourite music with Musiio here.
Register for our upcoming events:
- Join the Grand Finale of Intel Python HackFury2: 21st Oct, Bangalore
- WEBINAR: HOW TO BEGIN A CAREER IN DATA SCIENCE | 24th Oct
- Machine Learning Developers Summit 2020: 22-23rd Jan, Bangalore | 30-31st Jan, Hyderabad