How Dubdub.ai Is Using Its Emotional AI Stack To Retune The Voiceover Industry

The advent of GenAI has led to a groundbreaking shift in the realm of text-to-speech, propelling chatbots to greater sophistication and enriching experiences of users in the areas of learning and entertainment.

Spearheading this change is Bengaluru-based Dubdub.ai, which is one of the first few Indian startups focussed on building text-to-speech models for the entertainment industry. The startup provides audio professionals with the technology to generate voices with varied emotions.

Besides offering features like voice cloning, the AI dubbing platform allows its users to record video and audio clips in more than 50 languages.

At a time when there are concerns among industry players about GenAI making the jobs of voiceover artists redundant, Dubdub.ai claims to have built its thesis around creating a more collaborative environment where AI and human talent can work hand in hand.

With its GenAI capabilities, Dubdub.ai operates in the global text-to-speech market, which is projected to reach $7.6 Bn by 2029, driven by the rising need for Al-based tools and natural language processing.

What’s At The Core Of Dubdub.ai?

The startup was incorporated in 2021 by IIT Kanpur graduates Anubhav Singh, Rahul Garg, Rahul Sankhwar, and Anchal Jaiswal to cash in on the country’s growing vernacular content market. Soon the cofounders realised that there was more to the playing field as they could help recording studios and audio professionals address two of the biggest industry challenges — time and cost.  

Hence, Dubdub.ai was first launched as an audio and video dubbing platform using third-party GenAI models. Today, the platform offers its services in more than 50 languages, including Spanish, Portuguese, Italian, German, English, and Indian languages.

Speaking with Inc42, cofounder Singh said that since the startup works in the entertainment domain, it uses seven to eight separate models that run in series to help professionals in each stage of dubbing, including reviewing, editing, and paraphrasing.

Currently, Dubdub.ai is focussed on building a foundational model for text-to-speech generation. “We have built our own transformer-based text-to-speech model that is a state-of-the-art technology, capable enough to generate any emotion,” Singh said.

After spending extensive time with audio professionals and in recording studios, Dubdub.ai developed its models to introduce various voice modulations akin to directions given to voiceover artists by directors in a studio setting.

“We have an audio prompting module along with the text. Now you can give an audio prompt, and that audio will enable you to generate emotions in every aspect as humans can,” Singh explained, highlighting its scalability across multiple languages.

Dubdub.ai’s core text-to-speech technology currently enables users to produce audio output in over 20 voices, capturing diverse age groups, genders, and a spectrum of emotions, including anger and confidence. However, it is presently available only in Hindi and English.

The startup also collaborates with artists and secures licences to use their voices. This helps users obtain audio output in their preferred voice, including those of celebrity artists. 

The AI startup is backed by Accel Atoms, Waveform VC, Force Ventures, and several angel investors. Dubdub.ai has raised $1 Mn since its inception.

How Does Dubdub.ai Make Money?

Unlike most other text-to-speech platforms and AI dubbing companies, Dubdub.ai claims to be largely focussed on catering to big businesses in the entertainment industry and not individual social media content creators.

With its tech stack, the startup caters to entertainment companies that are looking to leverage the power of AI. Dubdub.ai claims to take their data to give high-quality professional studio output, which is 95% close to a human-voice recording output. 

Hence, the use cases of its offerings are across domains and range from providing regular voiceovers in multiple languages for audiobooks or radio shows to dubbing TV series, webinars, or promo videos.

Dubdub.ai has two main revenue generation streams – a self-serve model (users can sign up on the platform and use the tools) and a service-based model.

Self-serve users are charged a subscription fee between $30 and $100 per month. On the other hand, for the service-based model, which provides more hand-holding to the clients for optimal use of the platform’s technology, Dubdub.ai charges 5X to 10X more than its base subscription fee.

The startup enjoys a customer base of over 250 users. Around 15 enterprises, including MPL, SnapChat, Kuku FM, Gameskraft, Britannia, and MG, currently use Dubdub.ai’s services.

The Road Ahead For Dubdub.ai

In a market cramped with many similar solution providers, Dubdub.ai not only bets on its proposition of GenAI with emotional intelligence but also accuracy. Further, the startup claims that its AI models address the problem of hallucinations during the conversation process itself. It competes with the likes of Dubverse, DupDub, VisualDub, and many others.

While the cofounders did not reveal its current financial figures, Dubdub.ai aims to achieve an ARR of $100 Mn in the next two years. 

Besides, the startup is also looking at launching its text-to-speech model in at least six to seven languages. For this, the cofounders are looking to raise around $4 Mn in its next funding round.

Talking about GenAI taking over the jobs of voiceover artists, Singh said that platforms like Dubdub.ai would never replace humans but rather work in tandem to create more opportunities. 

As digital content creation maintains a steady uptrend, it will be interesting to see how Dubdub.ai’s flourishes in the entertainment industry with its GenAI-based voiceover tools.