This startup is helping ESL learners find their voice with AI

The inability to speak fluent English despite being highly educated is a common frustration shared by many students, job seekers, and professionals alike in India.

While most edtech tools focus on teaching the language, four students from IIT-BHU spotted a deeper problem—fluency, not literacy, which was holding people back.

Determined to change that, Akshay Akash and his college mates Anesh Srivastav, Ankit Kumar Pandey, and Akshat Baranwal founded Stimuler in 2022.  

It is a voice-first AI tutor that replicates real phone conversations and offers feedback to help users speak English fluently. The Bengaluru-based startup is leveraging advanced speech analysis and AI-driven learning to transform how people learn and practice spoken English.

“We realised most language learning tools help you learn a language. But with English, especially for career growth, you already know the language—you just can’t speak it fluently. That’s the real problem.”

From campus project to global platform

What started as a college project has quickly evolved into a full-fledged platform with global ambitions. Today, the entire product—from development to support—runs out of Bengaluru. 

But the team is building for a global user base that struggles with a deeply local problem: the fear of speaking English out loud. “It’s something we’ve all seen growing up,” Akshay says. “We just found a way to solve it—with AI, and a lot of patience.”

The roots of the founding team dates back to their college years. Akshay and Anesh were writing multiple newsletters in their third year of college, one of which was centered around education. 

That exploration triggered the idea of building something in the spoken English improvement space. With Akshay and Ankit's background in speech and debating, they brought a unique advantage in understanding the nuances of fluency and verbal communication. 

By the time they graduated in May 2023, their voice-first app had already acquired over 100,000 users and secured funding from a top-tier VC. 

Also Read

Her struggles with English led this Meghalaya woman to start a school in a remote village

In June 2023, they moved operations to Bengaluru.

“Ankit was the best AI person we knew in college, and also my debating partner. So, we had a lot of trust and synergy from the start. Akshat was the best app developer we knew—he had built multiple apps before, and we were certain we were going mobile-first. We had worked on projects like hackathons together, so there was strong mutual comfort and understanding,” says Akshay. 

How it works?

Unlike typical chatbot-based apps, Stimuler mimics a real phone conversation. Users can select topics such as hobbies, workplace small talk, or job interviews, and engage in a voice-only exchange with an AI bot. 

At the end of each session, they receive feedback across multiple parameters such as pronunciation, fluency, vocabulary, emotional tone, and more. 

Based on this analysis, the app recommends targeted exercises tailored to each learner’s weak areas such as pronunciation issues, vocabulary usage, fluency, emotional tone, and even energy levels. 

It then offers a highly detailed, nuanced understanding of the user’s spoken English. This is made possible through Stimuler’s proprietary in-house models trained to evaluate speech at the phoneme level, considering regional accents from India, Latin America, and Southeast Asia. 

After the call ends, the app generates a “speech report card” instantly that highlights areas for improvement. The app also recommends personalised practice exercises to help users improve specific weaknesses. 

According to the team, one of the key reasons users keep coming back to Stimuler is the quality of its feedback loop. 

“It’s simply brilliant,” says one user. “It’s a boon for introverts like me. Talking to an AI coach is much easier than talking to randos on the internet. The reports are very detailed and probably the best feature of this app. I’m in love with it.”

Competition and challenges

Some of Stimuler’s key competitors are Speak.com and Praktika, which operate in the global markets. However, what differentiates the product is its deeptech foundation. 

“Most general-purpose language models or assistants like Siri and Google Assistant are designed to understand your intent, not correct your speech,” explains Akshay. 

“Stimuler’s biggest strength is its in-house audio AI infra. Our app has the most comprehensive speech feedback engine in the world which allows us to tell users how exactly they are performing across 15+ speech metrics. The feedback engine also allows us to recommend exact exercises for people helping them improve their speaking skills objectively,” says Akshay.

To achieve meaningful feedback, the team had to build many of its models in-house—especially those that focus on phoneme-level analysis and accent-specific insights. 

The app runs on a pipeline that combines proprietary and open-source models to detect, analyse, and evaluate speech across various accents. Separate models assess pronunciation, vocabulary, and fluency, while another set dynamically generates personalized exercises for reinforcement. 

Speaking about challenges, Akshay says that one of the primary challenges for the team was building the core AI infrastructure that powers the voice-based product experience—an especially complex task given the relative nascency of Voice AI as a field. In addition to the tech hurdles, expanding and monetising the user base beyond India posed its own set of difficulties. 

“Setting up acquisition and monetisation pipelines in international markets has been an evolving and challenging task,” says Akshay. 

Despite these hurdles, the platform has made significant headway—today, 85% of its paying users are based outside India. 

Business and plans ahead

According to a report by Facts and Factors, the global online language learning market was worth $14.2 billion in 2021 and is expected to reach $28.5 billion by 2028 at a CAGR of 18.8%.

“India is still maturing as a monetisable market for us. But in 3 to 4 years, we believe it’ll become a key geography,” Akshay notes.

Stimuler operates on a freemium model. Free users get limited daily practice; premium plans start at $3/month depending on region, with annual options available.

The app is currently used by over 4 million people, with paying customers mainly in India, Latin America, and Indonesia.

While the company did not disclose its revenue numbers, Akshay mentioned that they have 60,000 paid users from 175 countries.

Stimuler first raised a $25,000 seed cheque from GradCapital in 2022 before officially launching the product. The startup recently raised $3.75 million in a pre-Series A round led by Lightspeed, SWC, and others.

The team said it plans to invest the latest funding in three core areas. First, it will expand product infrastructure by increasing compute capacity to support more intensive AI processing and real-time feedback generation. Second, the team intends to grow its lean engineering unit from four to around ten high-performing members to accelerate development. Finally, the company aims to scale user acquisition in key markets, especially across Latin America and Southeast Asia, with a strong focus on platforms like TikTok.

“Stimuler is solving a universal need—making spoken English proficiency more accessible with sharp execution and deep product insight. Akshay and his team are building a truly global company from India, with early traction across LATAM and SEA geographies validating both the opportunity as well as the approach,” says Harsha Kumar, Lightspeed Partner.

“As voice AI transforms learning, I believe Stimuler has built a platform to lead the category for the next billion users. We are excited to back their journey.” 

Stimuler is currently driven by a lean team of 10 full-time members, including its four founders. 

The long-term mission, as Akshay puts it, is “to ensure that no one in the world misses out on opportunities simply because they come from a non-native English-speaking background.” 

At a product level, the goal is to build a population-scale, AI-powered consumer platform that helps users rapidly progress from their current proficiency to a level of English fluency that’s respected globally.