If you love humming or just making random sounds while reading your newspaper or sipping a cup of coffee, Google's new machine learning Tone Transfer will surely excite you because it may help you create your own.
(Photo : Photo by David Ramos/Getty Images) BARCELONA, SPAIN - FEBRUARY 26: A logo sits illumintated outside the Google booth on day 2 of the GSMA Mobile World Congress 2019 on February 26, 2019 in Barcelona, Spain. The annual Mobile World Congress hosts some of the world's largest communications companies, with many unveiling their latest phones and wearables gadgets like foldable screens and the introduction of the 5G wireless networks.
You can easily access the new feature. All you need to do is go to its official website using your Android device, laptop, or desktop (Mac or Windows). Once you access the site, you can choose the "Let's play!" option to continue or tap the "Watch short video" option to see how it works.
You'll see different inputs, including acapella (singing), birds (chirping), Carnatic (singing), Cello (performing), pots and pans (clanging), or synthesizer (riffing). These options are already prepared, which you can quickly transform into different tunes, such as saxophone, flute, violin, or trumpet.
To use your own voice, you need to click the "Add your own" input option. This allows you to record your humming or any sound you create for 15 seconds.
Google's machine learning algorithm will change your voice into digital sound, transforming it into any of the given instrument tunes.
Before you start using Tone Transfer, here are important things you need to know
It is important to note that the quality of Tone Transfer's output will depend on your background noise and mic you're going to use to record your voice. To get a good recording, you might need to take a few tries before getting the output you want.
(Photo : Photo by Alexandre Schneider/Getty Images) SAO PAULO, BRAZIL - JUNE 25: A Baccarelli Institute student plays the violin during the distribution of food at the institute's headquarters in Heliopolis amidst the coronavirus pandemic on June 25, 2020 in Sao Paulo, Brazil. Instituto Baccarelli is a non-profit organization that teaches music to more than 1200 underprivileged and socially vulnerable children and young people in Heliopolis, one of the biggest favelas in Sao Paulo of over 200,000 inhabitants. The Institute is responsible for creating the first orchestra in the world that emerged in a favela, the Heliopolis Symphonic Orchestra. With the social and economic impact of the coronavirus (COVID-19) pandemic in these poor communities, Instituto Baccarelli decided to raise funds and the donations are used for food, cleaning and hygiene products.
Google's new technology uses the Differentiable Digital Signal Processing (DDSP) library created by the Magenta AI team, a company that focuses on developing open-source technologies to explore the use of machine learning in art.
With DDSP's help, Tone Transfer can use a neural network to change user audio input into DSP (Digital Signal Processors), converted into many instrument sounds.
For more news updates about Google's new machine learning technologies, always keep your tabs open here at TechTimes.