2024 Vertical Market Case Studies: Speech Technology in Entertainment

Article Featured Image

Indians all around the world love Bollywood and their Bollywood stars, but it has always been difficult for Bollywood fans to reach out to their favorite celebrities. That is all starting to change, as fans can now connect with some of their Bollywood idols and even win personalized video messages from them thanks to TrueFan and speech technologies from Resemble AI.

TrueFan built its business on the popularity of Bollywood, says Nimish Goel, founder and CEO of the company, which provides a mobile application to connect Bollywood celebrities and sports stars to their fans in India.

The key to TrueFan’s business is personalized videos, but because of the sheer size of the Indian market, each celebrity can’t record these videos again and again, given their time constraints. Even if a celebrity did have the time, for TrueFan to pay them for multiple recordings would be prohibitively expensive, a problem for the price-sensitive Indian market, with a per-capita income far below that of the United States, Goel adds.

The company knew it couldn’t do every video as an original, so it came up with a way to create a couple of structured templates and then use technology to fill in the rest or create additional videos at scale.

For example, the celebrity would say: “It’s your birthday” or “It’s Mothers’ Day.” The AI technology then fills in the requested name and a very short, personalized message. Using only a total of eight minutes of the celebrity’s voice, the AI can build the rest of the voice recording from the couple of templates that are provided. Competing technologies need as much as an hour of a celebrity’s recordings, which would be much more costly for TrueFan.

When TrueFan first started looking for technology to add the necessary voice elements at scale, speech AI was still in its infancy, Goel says, with most providers offering only basic speech synthesis. Resemble AI was different.

“I give full credit to their founder (Zohaib Ahmed); he understood our unique use case. We didn’t need long sentences,” Goel says.

The two companies also have similar philosophies regarding the use of AI-generated voices.

The celebrities are compensated based on the usage of their voices, and both TrueFan and Resemble AI have policies against using AI-generated voices without compensation.

“Our product relies on that consent piece,” Ahmed says. “We don’t care how [the voice] is being used, as long as it’s within the bounds of legal concerns.”

Resemble AI’s platform “is extremely locked down,” Ahmed says, referring to a technical design that prevents Resemble AI from using cloned voices without compensating the actors. To help protect against others stealing and using the voices, Resemble AI’s technology also includes audio watermarks to authenticate the company’s files.

Another factor in choosing Resemble AI was that its voice cloning technology could quickly handle large projects. TrueFan recently completed a project involving several thousand recordings in a single day.

Goel also cited the quality of the voice rendering from the Resemble AI technology. “They use their own proprietary models, which work very well for us. There are two elements in this speech synthesis. One is the quality of pronunciation. Second is the speaker familiarity—does it sound like me or like some other speaker?”

Other technology might have good pronunciation, but they tend to lack speaker familiarity, according to Goel, who also cited the pricing of the Resemble AI solution as another advantage.

Goel doesn’t hold back when discussing what Resemble AI has meant for TrueFan: “Without Resemble we wouldn’t have been able to take off our business,” he says. “If Resemble didn’t exist, we wouldn’t exist. I have no shame in admitting that. They were kind of pioneers for us. They were probably the only player in the market when we started with them that could provide the scale, the accuracy, the service, and the cost we needed.”

In the recent campaign in which thousands of videos were completed in a day, the client had a relationship with another speech tech provider, but TrueFan pushed using Resemble AI instead. Resemble AI’s accuracy was as high as 90 percent, whereas the competitor’s accuracy was only 55 percent, according to Goel. The technology’s accuracy is critical to the growth, according to Goel, who notes that while celebrities authorized AI generation of their voices, they still want to ensure that the synthetic voice sounds like them.

With Resemble AI, TrueFan has been able to grow its videos exponentially, which is the key to its growth in revenue and profitability, Goel says. Initially producing only 500 videos a month, the company is now producing 35,000 videos a month. In the past year alone, the company’s revenue has tripled, and it is now seven times higher than it was just a few years ago.

The improved scalability has also enabled TrueFan to add more celebrities to its base, going from 3 to 10 last year, and now to 42. “We’ve scaled our celebrities pretty aggressively, which is also part of the reason the videos have scaled up,” Goel says.

That success has attracted investors as well, with angel investors adding to the company’s funding, which has grown by 2.5 times.

Goel expects TrueFan to continue to expand its business, adding more celebrities and offering longer and more greeting types. It is also considering an expansion of its use of Resemble AI’s speech-to-text for other purposes. And as the cost of Resemble’s underlying hardware and software drops, that will further impact TrueFan’s profitability. 

Phillip Britt is a freelance writer based in the Chicago area. He can be reached at spenterprises1@comcast.net.

The Payoff

Since adding Resemble AI’s voice cloning technology to its Bollywood celebrity app, TrueFan has seen the following results:

  • Increased the number of videos produced each month from 500 to more than 35,000.
  • Enjoyed a 90 percent accuracy rate.
  • Increased its base of celebrities from 3 to 42.
  • Increased revenue by a factor of seven.
  • Attracted additional investors to increase its valuation by two and a half times.

SpeechTek Covers
for qualified subscribers
Subscribe Now Current Issue Past Issues