Foundational Literacy and Numeracy (FLN) has been identified as India’s most urgent education priority under the National Education Policy (NEP) 2020 and the NIPUN Bharat Mission. According to a report jointly developed by the World Bank and UNESCO Institute of Statistics, 53% of children in low- and middle-income countries cannot read and understand a simple story by the end of primary school.
The Challenge
In India’s multilingual, high-enrolment classrooms, assessing early reading at scale requires approaches that are efficient, consistent, and practical for teachers. Traditional oral reading assessments are resource-intensive and time-bound, making frequent, individual-level tracking difficult.
Our AI Solution
The AI- assisted Oral Reading Fluency (ORF) solution uses automatic speech recognition (ASR) to bring consistency and scale to early reading assessments, built using thousands of real classroom recordings of students in different languages from across India.
Teachers can assess reading fluency in under five minutes per student, compared to the ten minutes often required through manual methods.
By enabling smart grouping for remediation, ORF supports differentiated instruction and lets teachers focus more time on teaching.
In Rajasthan, census-based ORF assessments have led to a statewide 100-day remediation program in academic year 2025-26, launched by the Chief Minister, embedding reading assessments into the academic curriculum, followed by the endline assessments.
In Gujarat, the ORF data has informed structured assessment cycles, driving instructional reforms. The state-wide census-based ORF assessments were conducted between November and December 2025, with the next assessment cycle planned for Q4 2026.
How it Works
01
Audio Capture in Real Classrooms
A student reads a grade-level passage aloud on a teacher’s smartphone, optimized for low-end Android devices.
Audio Capture in Real Classrooms
A student reads a grade-level passage aloud on a teacher’s smartphone, optimized for low-end Android devices.
02
Audio Pre-Processing
Noise reduction and volume normalization ensure device or classroom sound variations don’t compromise transcription accuracy.
Audio Pre-Processing
Noise reduction and volume normalization ensure device or classroom sound variations don’t compromise transcription accuracy.
03
Speech-to-Text Conversion
ASR models trained on thousands of recordings handle child speech patterns.
Speech-to-Text Conversion
ASR models trained on thousands of recordings handle child speech patterns.
04
Text Alignment and Error Detection
A custom algorithm compares the transcript with the target paragraph, identifying substitutions, insertions, and deletions.
Text Alignment and Error Detection
A custom algorithm compares the transcript with the target paragraph, identifying substitutions, insertions, and deletions.
05
Fluency Scoring and Metrics
The system computes Correct Words Per Minute (CWPM), accuracy, fluency trends, and grade-level reading classification.
Fluency Scoring and Metrics
The system computes Correct Words Per Minute (CWPM), accuracy, fluency trends, and grade-level reading classification.
06
Timely Teacher Insights
Teachers receive clear, interpretable class performance with feedback on how to remediate children.
Timely Teacher Insights
Teachers receive clear, interpretable class performance with feedback on how to remediate children.
07
System-Level Reporting
Data aggregates into dashboards supporting FLN monitoring and state-level decision-making.
System-Level Reporting
Data aggregates into dashboards supporting FLN monitoring and state-level decision-making.
Where It's Deployed
Deployed in Rajasthan (integrated with the ShalaDarpan Shikshak App) and Gujarat (integrated with SwiftChat). Currently available in Gujarati, Hindi and English, with expansion to other languages planned.
2
States
Attribution and Impact
- Reduced Evaluation Time: The ORF assessment reduces evaluation time from about ten minutes to nearly five minutes per student, yielding cumulative savings of around five million teacher- hours across the two states.
- Consistent and Efficient Reading Evaluation: An evaluation study in Gujarat showed that about 90% of teachers found the ORF solution useful, especially for identifying learning gaps and grouping students by reading level. Additionally, 81% expressed willingness to continue using the solution, with perceived usefulness and ease of use as the strongest predictors of continued use.
- Data-Driven Prioritization of Learning Needs: In Rajasthan (academic year 2026-27), 64.7 percent of students in English and 34.3 percent in Hindi were identified as needing targeted remediation. In Gujarat, 69 percent of students were similarly identified, supporting data-driven prioritization of remediation efforts.
- In Rajasthan, following statewide remediation efforts implemented after the 2025–26 baseline assessment, endline assessment results indicate that approximately 40% of students in lower cohorts (C1 & C2) progressed to higher learning cohorts, highlighting notable progress in foundational reading levels.
- Economic Sustainability: With a cost of USD 0.00061 (about 5 paisa) per audio assessment, the solution demonstrates exceptional cost efficiency.
Our Other Solutions
SEAP
Read more →Partners
Looking Ahead
ORF offers a blueprint for how locally trained, responsibly deployed AI can support public education systems without bypassing teachers or institutions. By combining India-specific data, human-in-the-loop validation, and system integration, ORF demonstrates how AI can close capability gaps rather than widen them. The expansion of ORF is planned across Andhra Pradesh, Chattisgarh, Odisha, Meghalaya, Bihar, Uttrakhand and Bihar.