【2020 Solutions】 All 10,000 background music tracks are AI-made: Anpu targets the global broadcasting market
"Why does the music industry need AI? What problems can AI solve?" These are the questions Zeng Zhizhong, the founder and CEO of Anpu Artificial Intelligence Co., Ltd., continuously asks himself. Since founding the company in 2018, Zeng Zhizhong has understood market positioning and customer needs clearly. Anpu uses AI composition to seize the global environmental music broadcasting market. With the system online for over a month, AI has created more than 10,000 pieces of music, serving clients across cafes, car showrooms, restaurants, and hair salons with a "legal, free" broadcasting solution, becoming the first choice for stores.
Zeng Zhizhong is a serial entrepreneur with a background in both the internet and music. He has served as the general manager of Taihe Music Group, director of music services for Microsoft and Nokia in Asia Pacific, founded AR company Emitia Technologies and streaming company Tianlida Technology, and currently operates a company called Ouster Music specializing in film and television soundtracks.
With a dual background in technology and music, Zeng Zhizhong navigates AI music to solve copyright dilemmas
With his sharp senses in technology and music, particularly amidst the ongoing AI boom, Zeng Zhizhong constantly contemplates how to turn AI composition into a profitable business. He analyzes examples such as Spotify with 100 million paid subscribers in the USA, 1 million in Taiwan's KK BOX, and the mainland China's IPO-listed QQ Music, all operating at a loss. The primary issue lies with these platforms not owning the copyrights to the music they provide, despite offering membership subscriptions. They must also pay royalties to record companies and creators, leading to 'the bigger they are, the more they lose.'
Zeng Zhizhong, with his tech and music background, hopes to carve a niche for AI-generated music
In the music industry, there are two main domains: ambient music (background music, BGM) and pop music. Pop music involves a lengthy production chain including lyric writing, composition, arrangement, singing, harmonizing, mixing, and finalization, entailing high costs and investment risks. Meanwhile, ambient music, used in malls, department stores, cafes, and restaurants, traditionally sees copyrights held by music industry associations in various countries, making acquisitions costly and time-consuming. However, producing music in-house circumvents copyright issues. Thus, composing with AI and retaining copyrights internally becomes a key to success.
According to the International Federation of the Phonographic Industry (IFPI) report, global music market revenues in 2018 grew by 9.7% to $19.1 billion, up from $17.4 billion in 2017. Streaming music revenue alone reached $8.9 billion, accounting for 47% of the global total, nearly half. Publicly broadcasted music accounted for 10-15%, marking a significant portion of the market.
Recognizing the immense potential of the market, Zeng Zhizhong then assessed the technical capabilities of AI music, candidly stating, "AI is not a cure-all." For concert performances or chart-topping pop music, human lyricists and composers are necessary to achieve desired effects, while AI composition typically handles simple, uncomplicated melodies.
Assembling a large music database coupled with proprietary AI algorithms for rapid music production
Anpu Music's AI composition system utilizes algorithms that include Markov chains, neural networks, deep learning methods, and combines the company's proprietary algorithm MDN Music Deeplearning Network, which conforms to unique musical algorithmic theories, thus breaking through traditional pop music structures and styles to create more market-aligned music compositions. The database aggregates a large amount of sheet music data from top charted tracks and renowned songs globally, initially analyzing and summarizing the characteristics and melodies of popular quality music, then employing deep learning for efficient and excellent outcomes in AI composition.
▲BGMRADIO公播平台上集結上萬首AI音樂
Anpu provides a clear AI solution for the complex music copyright environment, with a material library owning a vast amount of clear-cut copyrights over 10,000 music tracks in 50 different styles, allowing users to freely choose suitable music to enjoy. Anpu's current business model is twofold: one provides a web-based platform offering 10,000 free AI-generated music tracks for online listening, and the other involves custom music services for a fee. Additionally, responding to the promotional needs of the record industry and artists, it also charges for advertisement playbacks. Another revenue model involves renting music players to users, charging an annual rental fee.
▲BGMRADIO公播平台與其他公播平台之比較
Zeng Zhizhong states, "Music knows no borders; good music doesn't distinguish between being created by humans or AI." With current AI algorithms and related technology being quite mature, using AI to produce music is not a difficult task. The key is identifying market pain points for business opportunities. Anpu's market spans Taiwan, Japan, Korea, Singapore, and it aims to continue expanding into China's largest market.
Having founded startups for 20 years in mainland China, Zeng Zhizhong's primary reason for starting a business back home is Taiwan's rich talent pool, especially the interdisciplinary talents. Unlike typical AI or music companies, Anpu requires a large number of amphibious talents capable of both programming and music. The company comprises two main departments: the R&D department, mostly formed from graduates from NTHU and NCTU in electrical engineering, electronic engineering, and applied music, and the music production department, where after quick AI algorithmic composition in the R&D department, highly musically educated producers refine these AI compositions into high-quality music experiences.
▲安譜團隊大多是科技與音樂兼具的跨域人才
「Translated content is generated by ChatGPT and is for reference only. Translation date:2024-05-19」