Sensibol builds and licenses solutions in collaboration with one of the premier audio research labs in India, the Digital Audio Processing Lab, IIT Bombay to actively pursue research in music analysis and have filed for multiple US patents for singing voice analysis. The company’s audio-processing algorithms, unique interactive singing interfaces has been deployed successfully across the music edutainment industry for Mobile Applications across Television, Telecommunication, Music Labels, Brands, and Music Schools.
Vishweshwara Rao Co-founder at Sensibol Audio Technologies Pvt. Ltd, talks about the sensible act of creating music education content platforms in the field of Indian Classical and western classical Music and reality shows to polish aspiring singers, developing Reading skills and making learning fun through gamification, thereby catering to a large segment of learners yet to be tapped in an interesting conversation with Loudest.in.
An Electrical engineer by Profession Vishweshwara Rao went to do his Masters in Music Technology from the University of Miami, Florida when he realized at a point that there was a lot of gap in terms of level of integration between technology and Indian classical music, while the western musicians used a lot of technology in their music, thus returned back to IIT Bombay to create an area in Music Technology.
Rao says, “In spite of being an electrical engineer, I was always interested in music and was curious to know where does music and technology meet. I came back to do a Ph.D. in music technology at IIT Bombay, wherein we created an area of Music Technology since it was not there as a subject initially but, Prof.Preeti Rao in the department of electrical engineering agreed to guide me who was researching on Speech. Eventually, we collaborated with NCPA in the research to identify what is it that the technology can help in Indian classical music under Dr.Suvarnalata Rao Programming Head NCPA, who became my co-guide.”
Explaining on the research process and the Inception of Sensibol, Rao said, “Initially we began by extracting the melody of a singer in the presence of the background music from a commercial track under ‘music information retrieval process, to build technology to extract information from music the way humans do. The process included identifying genre, tempo, and my work was on identifying Melody. Eventually, we submitted our work to a major technology competition in the US called MIREX ( Music Information Retrieval Evaluation eXchange) run by the University of Illinois at Urbana –Champaign. Fortunately, we came first amongst many research entries from across the world, which gave us a big boost following which IIT Bombay filed a patent to this technology. IIT Bombay has a business incubator wherein they help students to create startups and since the technology received good reviews me and my partner Sachin pant started up Sensibol.”
Talking from Indian Classical Music perspective and how the technology could help the learners Rao said, “In NCPA the interest was in classical music so they wanted to identify all the different kinds of Harkat’s (Improvisations) and variations to show a student how good or bad they sound and what is the scope of improvement. We analyzed some classical tracks of Hindustani classical exponents in 2009-2010 and also the classical exponent Mala Ramadorai, who teaches both Hindustani and Carnatic Music. We analyzed certain tracks and gave them representations in Hindustani music in which ‘Harkat’ are very subtle while on Carnatic music the ‘gamaks’ (Improvisations) are very deep, and so the students could perfect those improvisations by practicing.”
How the technology is being utilized by students Rao said, ” We have collaborated with the true school of music to make an app Learn to sing. The idea is to help the students learn from recordings itself if they are unable to have any formal training with a music teacher or a guru. The app is based on curriculums of both Hindustani Classical western music, and are in the process of developing a curriculum for Carnatic Music. The app has simplified learning grade-wise starting from Grade 1,2,3 which gets tougher as the grade increases. In Hindustani Classical Grade 1 has basic ragas, sargams and then gets tougher when you progress. The app helps to practice and listen to their Harkat’s (Intricate improvisations) and notes and students can improve them, as you can see every movement and slow it down to practice. Students can pick from a large catalog of contemporary to Classic English and Hindi songs, which will be updated regularly. So while we have a track of Kishore Kumar’s voice, so we could create a very detailed and a technical report on it to include his voice modulation, variations, style etc.”
Euphony is an online platform that enables music educators to provide content to and evaluate student musical proficiency. The Euphony App engages students to practice singing. Talking about the Singing guru evaluation technology being used in the app Rao said, “We found that there was a big gap in the teacher-student framework because students don’t practice, so every class just goes in revising the old lesson rather than teaching something new. This is a platform which helps students to practice and evaluate online.”
Rao elaborated on how does the Euphony App help in practicing Harmonies of western music, “Euphony is the platform that we have built for the churches, which is specifically designed for the choirs in the churches. We started up with the choirs, wherein a choir conductor can upload the choir music, and the choir singers download it and practice along with it. So each person gets one section Alto, Sopranos, Bass, Tenor and then they can practice on the app. So students can turn down the voices and the background music of the song and practice their own sections and figure out how well they are going, and when all come together can do a harmony together.”
So how does the technology works, Rao describes, “When you sing in the interface you get a reflection over the intricacies and can then improvise. Thus, the choir conductor gets to know everyone’s performance as he gets all the data. So the teacher can mark on every line of the composition to tell what went right and what went wrong and tell them to make it perfect.”
How does the technology Incorporate Machine learning and AI Rao said, “Since music is rule-based so is the system. Thus, to judge an improvisation, we use machine learning based on human inputs data, which tells that how an improvisation make people say wow, so that is the AI part, at the moment it is more of singing processing and music algorithms at the moment.”
Talking about key clients in the music industry, “We made a singing platform called ‘Gaona’ then and showed the technology to HMV Saregama. They got interested and licensed it for their apps which they were making way back in 2014, that they were building, so it was a good learning experience for us in associating with a big brand then.”
Rao Adds, “ We did a project called ‘ Airtel Trace Music Stars’ with Airtel in Africa across 13 countries. So a caller would call a number and sing a song and would get scores on the spot. So each country had its own contest and the winner from across countries got a chance to be mentored by the singer Akon. Eventually, we launched karaoke apps in Afghanistan, Bangladesh, Sri Lanka, which were for Telco’s like Etisalat karaoke in Afghanistan, Mobitel in Srilanka.”
Talking about using the technology helping the reality shows “We have partnered with Sa Re Ga Ma Pa Lil Champs on Zee TV to create its digital version, in which kids come and see how they are performing and submit their recordings, and get a chance to feature on the show and get a prize, so this is how the awareness about the technology is helping children sing better.”
Rao talks about how digital apps can help judge music performances, “Digital cannot judge like the humans since music is very subjective, but can help in analyzing and shortlisting bulk of tracks to help humans. We did a project with Universal Music Group for Max Life Insurance ‘Young Singing stars’. It was a contest for children between 9-14 years of age, which had on-call auditions for shortlisting out of 50000-75000 entries, which they could not do manually and used our technology to shortlist.
Rao Adds, “ We did an audit and took some random tracks and gave it to a human judge, which had a good score on our system as the child sang well after initial 5 seconds, but the judge rejected it. So human judges do have a fatigue level and so they judge in the first 5 seconds and if they don’t like it they reject it, while in the following seconds the child can sing extremely well. So that’s where the humans have a limitation when it comes to bulk and the digital can help bring out the right talent. Also, the system helps detect auto-tuning and cheating, which humans sometimes miss and the machine picks it.”
Talking about working with digital OTT platforms to engage users Rao said, “We are diversifying into Gamification across TV, video platforms like we did one for ZEE5 as we developed for online ‘Guess the score’ where the singer finishes singing and gets 60 sec to judge what score the jury will give and if the score matches with the jury then you win a prize, so it was to engage users.”
Talking on the need to build educational apps Rao said, “We are also building something called a ‘reading tutor’ so while reading a comprehension, it guides to polish the speech, pronunciation and that’s a key part of how someone presents himself, so these are very important skills which should be taught from an early age. To understand gaps, pauses, rhythm, pronunciation, intonation it’s being used for experimentation in a school in Himachal by IIT Mumbai.”
Talking about partnerships and opportunities Rao adds, “Our App Gaona is a platform which we call Tiktok meets Byju’s, which is music-centric where the quality of music is very important and also has an education piece in it and we have a platform ready and are looking for partners to promote it. We see a big opportunity in this space as by using this technology the music industry can be reached to interiors of India and get talent from there too.”
Concluding with his Music Inc. experience Rao said, “It was a great platform in terms of how it gave a direct interface to the music industry, musicians and artists thereby giving them an actual contact point. Also, I feel that a lot of musicians want to learn the art and came across a massive demand for learning content, especially in Indian classical music, so we have a lot of scope for building music educational videos and content, to teach people an art. Every musician must look at teaching music as an avenue from the digital perspective. So if someone wants to learn one must go to True school which today has 200,000 learners on the app. There is a massive demand for learning which Sensibol is suffixing.”