Soket AI Labs Launches Realtime Speech API, Offers Multilingual Capabilities

Soket AI Labs, the Gurugram-based AI startup, has unveiled its Realtime Speech API, aiming to transform AI interactions with voice intelligence and seamless integration. 

Soket AI Labs claims that the Realtime Speech API boasts ultra-low latency of under 500 milliseconds, ensuring near-instantaneous responses for real-time interactions. It supports multilingual capabilities to overcome language barriers and includes advanced functionalities such as tool calling, Retrieval-Augmented Generation (RAG) support, custom voice creation and cloning, and the ability to handle dynamic voice interruptions for natural conversations.

Developers can integrate the API effortlessly within 1-4 weeks using SDKs available for Python and JavaScript. The service is priced competitively at $0.012 per minute, providing an affordable alternative to industry leaders like OpenAI.

Soket AI Labs emphasised the platform’s versatility, highlighting its applications across industries such as banking, financial services, insurance (BFSI), healthcare, and telecommunications. Additional features include fine-tunable models and customisable voice options to meet specific business needs.

The company is set to launch its “Voice Innovators Beta Program” soon, inviting users to explore and shape the future of voice technology.

In a separate post on LinkedIn, Abhishek Upperwal, founder and CEO of Soket AI Labs, highlighted the importance of making ‘General Voice Intelligence’. “Voice is one of the most important interfaces to AI today and language models are at the core of intelligence,” Upperwal said.

In May, the company also launched India’s first open source multilingual foundational model, Pragna-1B. Upperwal said that it took the company six months to train the model, which involved many experiments with different models and a total of 150 billion tokens.

Founded in 2019, Soket AI Labs’ focus was on building a decentralised data exchange for smart cities. However, things changed significantly after OpenAI CEO Sam Altman’s visit to India, which motivated the team to build the best AI models in the country. 

Apart from Soket AI Labs, startups like Sarvam AI and CoRover.ai have been focused heavily on building speech models. Speaking at Cypher 2024, Sarvam AI chief Vivek Raghavan demoed the speech capabilities of its AI models, leaving everyone at Cypher speechless.