PolyWhisper
v1.0Apache 2.0Indic Multilingual Speech Recognition
PolyWhisper v1.0 is a indic multilingual speech recognition. Operates across Base: Whisper-Small (frozen) · Adapters: Dual LoRA Experts · Router: 33K parameter selector · Languages: Hindi · Tamil · Telugu · Bengali · Marathi · Format: ONNX · HuggingFace · License: Apache 2.0. License: Apache 2.0.
Indic multilingual speech recognition with frozen Whisper encoder and language-specific LoRA experts. Supports Hindi, Tamil, Telugu, Bengali, and Marathi with 20× hallucination reduction. ONNX quantized for edge deployment.
- ✓20× reduction in ASR hallucination on Indic languages
- ✓Frozen encoder preserves multilingual generality
- ✓ONNX quantised for edge and mobile deployment
- ✓Handles Hinglish code-switching natively
Cite this Model & Architecture
If you utilize PolyWhisper in your academic research, benchmarks, or enterprise deployments, please cite our open-weight publication:
@misc{eulogik2026polywhisper,
title = {PolyWhisper v1.0: Indic Multilingual Speech Recognition},
author = {Kishore, Gautam and Eulogik Systems Engineering},
year = {2026},
publisher = {Eulogik},
howpublished = {\url{https://eulogik.com/models/polywhisper}}
}Need help deploying PolyWhisper?
We offer consulting, custom fine-tuning, and on-premise deployment for all our models.