Mumbai based startup SensiBol has a developed technology that will interest all audiophiles and those in the music industry. The company’s software tells users just how good or bad they are at singing. It also helps them improve their singing.
Founded by Vishweshwara Rao, Sachin Pant and Professor Preeti Rao of IIT-B, SensiBol started out as a student research project. Vishweshwara Rao was working on his PhD in audio processing when he developed the software.
“Many companies were asking for audio processing technologies from us. We realised that there is much potential in developing one,” says Vishweshwara, who raised funding from the Indian Angel Network last year to grow SensiBol.
The IIT-B incubated startup had filed and been granted a US patent as well.
Singing Voice Processing
The GaoNa Karaoke App showcasing SensiBol’s Technology
The singing voice processing(SVP) technology has four parts to it.
Singing rating evaluates singing quality by comparing it with the original singer’s voice on dimensions like pitch, modulation, lyric pronunciation and octave change among others. It also gives users real-time visual feedback.
Lead melody suppression systems in the SVP creates karaoke like music tracks for singers to sing along with
Effects mixing with processing options that range from mono to virtual studio recording and,
Real time acoustic feature extraction with octave-error correction and silence detection among other features.
Audio Content Analysis
On top of the analysis of voice quality, SensiBol also offers services that extract information from audio.
Music meta-data extraction in which SensiBol’s algorithms extract tonal, timbral and structural information.
Music similarity analysis algorithms that identify the similarities between music tracks and help locate duplicates, and,
Speech analysis that gives users real time feedback on speech and spoken language utterances
SensiBol licenses the technology out to companies and the commercial model involves a license fee per platform. They also customise the licenses and prices according to the needs of the clients.
“If there is a singing reality show which wants to license our technology, then we customize and license the technology to them. People can also build apps on top of our APIs,” explains Vishweshwara, who has already signed up two enterprise customers. The startup is also talking to prospects including music labels, film production houses and telcos.
The SensiBol app, a consumer avatar of the technology, has been downloaded 2000 times. “On an average, people engage with the app for about 18 minutes,” he said.
The company will release APIs for meta-data information soon. They plan to broadbase their technology in the future in other areas of audio technology apart from music like spoken language evaluation for recruitment and BPOs.
“We want to become the intel for audio technology where we are such a potent technology company which can plug tech across use cases which span music, speech, general audio and others,” explains Vishweshwara.
The startup with a 10 member team, is also planning to tie up with hardware companies for better distribution.
The GaoNa App
If you want to take a look at how exactly the technology works, try SensiBol’s GaoNa app. The karaoke app is a B2C offering from SensiBol that is used only to showcase the use of the singing and voice processing technology. It has 1,000-5,000 downloads on the Play Store. Get it here.