Fact · Languages and accents
A 2026 study of open-source speech recognition models on Indian speech analyzes Hindi, Bengali, Kannada, Telugu and Marathi.
Checked against its source on October 17, 2026
In the source's own words
“We analyze linguistic, speaker-level, and acoustic factors across Hindi, Bengali, Kannada, Telugu, and Marathi.”
Factors affecting ASR performance: A study using state of the art ASR models in Indic Languages, Agneedh Basu et al., arXiv preprint, 2026, Abstract.
Open the original- What it measures
- The study looks at why recognition varies, such as speaking rate, telephone codecs and background noise, not only how much it varies.
- What it does not show
- The study is zero-shot (no retraining) and uses open-source models only. It is a preprint, so its results may change in peer review.