listen-windy-lingua-ml

WindyWord.ai STT — Malayalam Lingua (GPU (safetensors))

Part of the Windstorm Labs open model catalogue. Scores, licence and attribution for every model: https://windytranslate.com/models/listen-windy-lingua-ml

Transcribes Malayalam speech (Dravidian > Southern Dravidian).

Note: Quality ceiling: audited at 73.3% WER (community Malayalam Whisper space is thin — best alternative on HuggingFace audited at 76.5%, ~1.5% worse). Source: vrclc/Whisper-small-Malayalam. For high-stakes Malayalam transcription consider openai/whisper-large-v3 multilingual.

Quality

• FLEURS WER: 73.2% (50-sample audit)
• CER: 0.4872
• Band: unusable-gap
• Source: our 50-sample FLEURS screening (not a publication result)

About this variant

This is the safetensors deployment format of our Malayalam Lingua STT model. Load it via the safetensors/ subfolder.

Part of the WindyWord.ai STT fleet — covering 35+ languages that commercial speech-to-text APIs underserve, with proper dialect / script disclosures where they matter.

Usage

from transformers import WhisperForConditionalGeneration, WhisperProcessor
processor = WhisperProcessor.from_pretrained("WindyWord/listen-windy-lingua-ml", subfolder="safetensors")
model = WhisperForConditionalGeneration.from_pretrained("WindyWord/listen-windy-lingua-ml", subfolder="safetensors")

Apps

The Windy Word apps are built on this model family.

---

Provenance & License

Weights derived from vrclc/Whisper-small-Malayalam (community fine-tune by vrclc; identified from this repository's commit history and config.json). Licensed Apache-2.0 upstream; this redistribution is released under the same licence.