whisper-base-ar-quran (ggml, q5_1)

GGML version of tarteel-ai/whisper-base-ar-quran for whisper.cpp and whisper.rn. The Quranic Wird app uses it to follow a recitation on the phone.

Changes from the original model

Converted with models/convert-h5-to-ggml.py from whisper.cpp. Before the conversion, max_length in config.json was set to max_target_positions (448), otherwise whisper.cpp rejects the model.
Quantized to q5_1 with whisper-quantize.

The weights were not retrained or fine-tuned.

File	Size	SHA-256
`ggml-tarteel-whisper-base-ar-quran-q5_1.bin`	59.7 MB	`acc1f03c8b1468b674b4fab50e0b2312cff982ba0001cd16fe36ca23862b6536`

Apache-2.0, like the original model. The original weights are the work of Tarteel AI.

Downloads last month: -; Downloads are not tracked for this model. How to track

Base model

Finetuned

(16)

this model