Blue Machines AI Unveils Aurora Speech-to-Text Model Built for India's BFSI Sector

New Delhi: Enterprise AI platform Blue Machines AI has launched Aurora, a multilingual speech-to-text model designed specifically for India's banking, financial services and insurance (BFSI) sector.

The model, launched on September 7, is designed for real-time financial conversations in which customers may switch between languages, combine English financial terminology with regional languages or communicate over noisy and low-bandwidth telephone connections.

According to Blue Machines AI, Aurora was evaluated internally on representative BFSI datasets covering banking, lending, insurance, collections and customer-service interactions. The model recorded a Semantic Word Error Rate (WER) of 1.51% for English, 2.43% for Hindi BFSI conversations and 5.52% across multilingual speech.

Aurora also recorded a BFSI Entity Error Rate of 4.23% for information such as monetary amounts, interest rates, policy numbers, account references and transaction IDs. The company said it compared the model with leading speech-to-text systems using consistent audio inputs and scoring methodologies.

Built for India's Multilingual Financial Conversations

Unlike general-purpose speech recognition systems, Aurora has been trained to understand terminology commonly used in financial services, including EMIs, outstanding amounts, foreclosure charges, disbursals, KYC, premiums, SIPs, NAVs, policy numbers and transaction IDs.

The model is also designed to recognise numbers, currencies, percentages and financial identifiers that are critical for downstream banking and insurance workflows.

The datasets used for evaluation included Indian English, Hindi, Hinglish, code-mixed speech, regional pronunciation patterns, background noise and telephony audio. This focus is intended to address the linguistic complexity of India's financial customer interactions.

High-Concurrency Real-Time Processing

Aurora uses a cache-aware FastConformer encoder and streaming transducer decoder, allowing it to process speech incrementally while retaining conversational context, according to Blue Machines AI Chief Technology Officer Abhishek Ranjan.

In internal throughput tests, the model supported up to 960 concurrent real-time streams per Nvidia H100 GPU at a 320-millisecond operating point, while throughput increased to 2,400 concurrent streams per H100 at a 1.12-second operating point.

The company has also developed training pipelines that allow Aurora to be adapted using customer-authorised enterprise data. This enables financial institutions to customise the model for proprietary product names, terminology, geographies, accents and interaction patterns.

Blue Machines AI said customer-specific retraining produced a 40-45% relative reduction in recognition errors compared with the base model on institution-specific datasets.

Designed for Banking and Insurance Workflows

Aurora is integrated with Blue Machines AI's enterprise customer-experience AI platform and can be deployed across multiple financial customer journeys.

Potential applications include customer acquisition, onboarding, lending, collections, servicing, insurance, claims and customer support.

The model can be deployed through managed cloud infrastructure, an enterprise virtual private cloud or on-premises environments. This flexibility is intended to help financial institutions meet their requirements around data security, data residency and governance.

Blue Machines AI founder and CEO Nirmit Parikh said India's financial conversations often do not follow a single language or standard script. The company has positioned Aurora as part of its broader effort to build sovereign AI infrastructure for Indian enterprises.

The launch comes shortly after Blue Machines AI announced Project Icebreaker, a sovereign AI co-innovation programme under which the company plans to work with five Indian financial institutions to move customer-experience AI initiatives from proof-of-concept stages into production.

Focus on Sovereign AI for BFSI

The emphasis on sovereign AI is particularly relevant to the BFSI sector, where institutions handle sensitive financial and personal information.

Blue Machines AI's platform supports deployment in enterprise virtual private clouds, private cloud environments and on-premises infrastructure, allowing institutions to retain greater control over data, models, workflows and governance.

With Aurora, the company is seeking to address one of the key challenges in India's AI adoption journey: enabling speech-based systems to understand the country's multilingual and code-mixed financial conversations accurately enough for high-stakes enterprise use.