Bengaluru is slowing turning into a key hub for artificial intelligence in India, and Sarvam AI is the latest homegrown startup from the city to grab attention. Founded in 2023, Sarvam AI has taken a different route from most global AI companies. From the start, it has focused on Indian languages, Indian users, and problems that are common across the country. Built entirely in India, the company’s main aim is to support India’s push toward self-reliance in AI.
SurveySarvam AI positions itself as a strong alternative to global models like OpenAI’s ChatGPT and Google’s Gemini when it comes to Indian language understanding. I personally compared Sarvam with ChatGPT and Gemini across three key areas (text-to-speech, speech-to-text, and translation) to see if it really lives up to the hype.
Text-to-speech
I used the same video script for all tools. Sarvam AI offered a wide range of voice options, all with Indian accents. Some were subtle, others more pronounced. Choosing the right voice took time, but the result was impressive. The selected voice sounded natural, used proper pauses, and even included small fillers like “uh,” which made it feel human rather than robotic.
ChatGPT did convert the text to speech, but the output had issues. The audio echoed and broke at several points, making it hard to listen. In Gemini’s case, direct text-to-speech was not available, so I tried Google’s NotebookLM. That only gave a summary-style audio, not a full reading of the script.
Also read: OpenAI co-founder says agentic engineering is the next big thing in AI coding
Speech-to-text
For transcription, I uploaded an interview recording. Sarvam AI handled the audio well and gave an accurate transcript. ChatGPT did not transcribe the audio and instead suggested using another transcription tool and then pasting the text for cleanup. Meanwhile, Gemini initially rejected the file, saying it was too long. After trimming the audio, it worked, but breaking one interview into multiple parts is not a practical solution.
Translation
Translation is where Sarvam AI struggled. I provided a Telugu news paragraph, and the translated output contained factual mistakes. In contrast, both ChatGPT and Gemini translated the same content smoothly and accurately, without losing the meaning.
Verdict
Sarvam AI stands out for voice-based tasks, offering natural Indian-accented text-to-speech and reliable speech-to-text. However, for translation and overall language accuracy, ChatGPT and Gemini still perform better.
Follow Us
Ayushi Jain
Ayushi works as Chief Copy Editor at Digit, covering everything from breaking tech news to in-depth smartphone reviews. Prior to Digit, she was part of the editorial team at IANS. View Full Profile
